Greenwich Academic Literature Archive (GALA)ndash the University of Greenwich open access repository
httpgalagreacuk
__________________________________________________________________________________________
Citation for published version
McCarthy Elizabeth W Arnold Sarah EJ Chittka Lars Le Comber Steven C Verity Robert Dodsworth Steven Knapp Sandra Kelly Laura J Chase Mark W Baldwin Ian T Kovařiacutek Aleš Mhiri Corinne Taylor Lin and Leitch Andrew R (2015) The effect of polyploidy and hybridization on the evolution of floral colour in Nicotiana (Solanaceae) Annals of Botany 115 (7) pp 1117-1131 ISSN 0305-7364 (Print) 1095-8290 (Online) (doi101093aobmcv048)
Publisherrsquos version available at
httpdxdoiorg101093aobmcv048
__________________________________________________________________________________________
Please note that where the full text version provided on GALA is not the final published
version the version made available will be the most up-to-date full-text (post-print) version as
provided by the author(s) Where possible or if citing it is recommended that the publisherrsquos
(definitive) version be consulted to ensure any subsequent changes to the text are noted
Citation for this version held on GALA
McCarthy Elizabeth W Arnold Sarah EJ Chittka Lars Le Comber Steven C Verity Robert Dodsworth Steven Knapp Sandra Kelly Laura J Chase Mark W Baldwin Ian T Kovařiacutek Aleš Mhiri Corinne Taylor Lin and Leitch Andrew R (2015) The effect of polyploidy and hybridization on the evolution of floral colour in Nicotiana (Solanaceae) London Greenwich Academic Literature ArchiveAvailable at httpgalagreacuk13251
__________________________________________________________________________________________
Contact galagreacuk
The effect of polyploidy and hybridization on the evolution of floral colourin Nicotiana (Solanaceae)
Elizabeth W McCarthy1ndash3y Sarah E J Arnold1z Lars Chittka1 Steven C Le Comber1 Robert Verity1xSteven Dodsworth13 Sandra Knapp2 Laura J Kelly13 Mark W Chase3 Ian T Baldwin4 Ales Kovarık5
Corinne Mhiri6 Lin Taylor7 and Andrew R Leitch11School of Biological and Chemical Sciences Queen Mary University of London Mile End Road London E1 4NS UK
2Natural History Museum London SW7 5BD UK 3Jodrell Laboratory Royal Botanic Gardens Kew Richmond Surrey TW93DS UK 4Max Planck Institute for Chemical Ecology Department of Molecular Ecology Beutenberg Campus Hans-Knoll-Strasse 8 07745 Jena Germany 5Institute of Biophysics Academy of Sciences of the Czech Republic CZ-61265 Brno CzechRepublic 6Institut Jean-Pierre Bourgin UMR1318 INRA-AgroParisTech INRA-Versailles 78026 Versailles cedex France
and 7Department of Plant Sciences University of Cambridge Downing Street Cambridge CB2 3EA UK For correspondence E-mail arleitchqmulacuk
dagger Present address Department of Botany and Plant Sciences University of California Riverside 900 University AvenueRiverside CA 92521 USA
Dagger Present address Natural Resources Institute University of Greenwich Chatham Maritime Kent ME4 4TB UKx Present address MRC Centre for Outbreak Analysis and Modelling Department of Infectious Disease Epidemiology
Imperial College London London W2 1NY UK
Received 6 October 2014 Returned for revision 15 January 2015 Accepted 16 March 2015
Background and Aims Speciation in angiosperms can be accompanied by changes in floral colour that mayinfluence pollinator preference and reproductive isolation This study investigates whether changes in floral colourcan accompany polyploid and homoploid hybridization important processes in angiosperm evolution Methods Spectral reflectance of corolla tissue was examined for 60 Nicotiana (Solanaceae) accessions (41 taxa)based on spectral shape (corresponding to pigmentation) as well as bee and hummingbird colour perception in orderto assess patterns of floral colour evolution Polyploid and homoploid hybrid spectra were compared with those oftheir progenitors to evaluate whether hybridization has resulted in floral colour shifts Key Results Floral colour categories in Nicotiana seem to have arisen multiple times independently during theevolution of the genus Most younger polyploids displayed an unexpected floral colour considering those of theirprogenitors in the colour perception of at least one pollinator type whereas older polyploids tended to resembleone or both of their progenitors Conclusions Floral colour evolution in Nicotiana is weakly constrained by phylogeny and colour shifts do occurin association with both polyploid and homoploid hybrid divergence Transgressive floral colour in N tabacum hasarisen by inheritance of anthocyanin pigmentation from its paternal progenitor while having a plastid phenotypelike its maternal progenitor Potentially floral colour evolution has been driven by or resulted in pollinator shiftsHowever those polyploids that are not sympatric (on a regional scale) with their progenitor lineages are typicallynot divergent in floral colour from them perhaps because of a lack of competition for pollinators
Key words Evolution floral colour hybridization Nicotiana flower pigmentation pollinator shifts polyploidySolanaceae spectral reflectance transgressive traits
INTRODUCTION
Polyploidy or whole-genome multiplication has played animportant role in the evolution of flowering plants (Soltis et al2009 2014) Allopolyploidy arising from interspecifichybridization and polyploidy can cause lsquogenomic shockrsquo(McClintock 1984) which may trigger a suite of geneticchanges including (retro)transposition differential gene ex-pression chromosome rearrangements and epigenetic changes(Leitch and Leitch 2008) These events and novel cisndashtransinteractions between progenitor genomes may generate varia-tion including transgressive phenotypes and facilitate rapid
divergence of both homoploid and allopolyploid hybrids(Wittkopp et al 2004 Chen 2007 Gaeta et al 2007 Anssouret al 2009 Tirosh et al 2009 Clare et al 2013)Speciation in angiosperms can be accompanied by or per-
haps driven by changes in floral colour that may influence pol-linator preference and reproductive isolation Many pollinatorssuch as bumblebees and hummingbirds visit a range of flowercolours (Waser et al 1996) Several species of flower-naivebumblebees have an innate colour preference for violet andblue shades although preferences in experienced foragers arelargely determined by learned associations between colours and
VC The Author 2015 Published by Oxford University Press on behalf of the Annals of Botany CompanyThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicenses
by40) which permits unrestricted reuse distribution and reproduction in any medium provided the original work is properly cited
Annals of Botany 115 1117ndash1131 2015doi101093aobmcv048 available online at wwwaoboxfordjournalsorg
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
rewards (Raine et al 2006) Hummingbirds appear to have noinnate preferences for particular colours but are likewise goodat forming associations between flower visual displays and re-wards (Goldsmith and Goldsmith 1979 Chittka and Waser1997) Hummingbirds have red receptors whereas many insectsdo not Consequently red flowers are poorly detectable by beepollinators but conspicuous for hummingbirds Thereforehummingbird-visited flowers are often red whereas those polli-nated by bees typically have a range of other colours(Rodriguez-Girones and Santamaria 2004 Shrestha et al2013) Flowers visited by nocturnal pollinators are more oftenwhite than those pollinated in full daylight probably to maxi-mize their detectability in dim light conditions (Kevan et al1996) Because of such differences in affinities of various polli-nator classes to certain flower colours differences in flowercolour can contribute to restricting gene flow between pheno-types although they will rarely result in complete reproductiveisolation for this differences in multiple traits are typically es-sential In Aquilegia (Ranunculaceae) blue- red- and whiteyellow-flowered species are primarily pollinated by bees hum-mingbirds and hawkmoths respectively (Grant 1952 Whittalland Hodges 2007) In Petunia axillaris (Solanaceae) hawk-moths prefer white flowers to pink flowers transformed to ex-press ANTHOCYANIN2 whereas bumblebees prefer pinkANTHOCYANIN2 flowers to white flowers demonstrating thatexpression of a single gene can cause differences in pollinatorvisitation (Hoballah et al 2007) Similarly manipulation of asingle locus controlling carotenoid production inMimulus flow-ers (Phrymaceae) results in a pollinator shift reaffirming theimportance of floral colour in determining pollinator behaviour(Bradshaw and Schemske 2003)To analyse floral colour in the context of pollination it is
necessary to consider both colour theory and pollinator visualsystems There are several important differences between thecolour vision systems of humans and those of various pollinatortypes Humans and many insects are trichromatic havingthree discrete photoreceptor types however humans possessred- (with peak sensitivity (kmax) near 560 nm) green-(kmaxfrac14 535 nm) and blue-sensitive (kmaxfrac14 420 nm) photore-ceptors (Bowmaker and Dartnall 1980) whereas many insectshave ultraviolet- (UV- kmax 350 nm) blue- (kmax 440 nm)and green-sensitive (kmax 530 nm) receptors (Peitsch et al1992 Briscoe and Chittka 2001 Kelber et al 2003) Manybirds (Bowmaker 1998) and some butterflies (Kelber 2001)have tetrachromatic colour vision In birds photoreceptors aresensitive to red green blue and either violet or UV wave-lengths (Hart and Hunt 2007) Hummingbirds have four singlecone types with peak sensitivities in the UV (kmaxfrac14 370 nm)blue (kmaxfrac14 440 nm) bluendashgreen (kmaxfrac14 508 nm) and yellow(kmaxfrac14 560 nm) the sensitivity of the last extends significantlyinto the red (Herrera et al 2008) Hummingbirds can learn todistinguish near-UV light (370 nm) from darkness confirmingthat they use their UV receptors for colour vision at a behaviou-ral level (Goldsmith 1980) We will take into account these dif-ferences in pollinator perception as we examine floral colour inthe genus Nicotiana (Solanaceae)We investigate the evolution of floral colour across
Nicotiana (Solanaceae) in the context of polyploidy and hybrid-ization Nicotiana is an excellent group in which to study the ef-fects of hybridization as nearly half of the 76 species are
allotetraploids of different ages (Chase et al 2003Clarkson et al 2004 2005 Leitch et al 2008 Kelly et al2013) and several putative homoploid (diploid) hybrids havealso been detected (Clarkson et al 2010 Kelly et al 2010)which add to the reticulate nature of the genus These phyloge-netic studies have also been used to predict the closest livingdescendent species of the parents that formed the homoploidhybrid and allopolyploid species hereafter called progenitorspecies as shown in Fig 1 Some synthetic polyploidsmade from these progenitor species are also availableproviding insight into the immediate effects of polyploidy andhybridization We compare floral colours of Nicotianapolyploid and homoploid hybrids with those of their diploidprogenitorsBecause various animal groups have different sensitivities
to colour it is necessary to model colour perception ofspecific pollinator classes to understand the significance offloral colour signals Here we consider floral colours from abee perspective (Chittka 1992) which can also be used to rep-resent other trichromatic insects such as hawkmoths due tosimilarities in photoreceptor sensitivities (Kelber et al 2003)and a hummingbird perspective (Herrera et al 2008 Restrepo2013) Hummingbirds and hawkmoths are known tovisit Nicotiana species (Aigner and Scott 2002 Kaczorowskiet al 2005 Kessler and Baldwin 2006 Nattero and Cocucci2007)Pigments typically determine floral colour however few
studies have examined the specific pigments present inNicotiana petals Aharoni et al (2001) confirm the presence ofanthocyanin pigmentation in N tabacum which seems to bepredominantly cyanidin derivatives Spectral colour shifts canoccur in anthocyanins due to hydroxylation and methylationwhich result in different types of anthocyanins (Castaneda-Ovando et al 2009 Andersen and Jordheim 2010) and differ-ences in pH as well as copigmentation with other pigmentsincluding carotenoids and colourless flavonoids or metal ionscan also cause spectral shifts in the same anthocyanincompound (Grotewold 2006 Andersen and Jordheim 2010)The yellow flower colour of Nicotiana glauca is due to caroten-oid pigmentation (Zhu et al 2007) Crossing experimentsbetween diploid Nicotiana species suggest that the presence ofchlorophyll in corolla tissue is dominant (Brieger 1935)and similar results corroborate this in the carnation Dianthuscaryophyllus (Caryophyllaceae Ohmiya et al 2014)In this paper we seek to determine what types of spectral
reflectance are found within Nicotiana and how they appear tobee and hummingbird pollinators We focus on the conse-quences of polyploidy and interspecific hybridization on floralcolour evolution Specifically we aimed to test the hypothesesthat (1) polyploid and homoploid hybrids will have floralcolours that will resemble at least one of their progenitors indiscrete spectral bee and hummingbird floral colour categoriesobtained from cluster analyses (2) polyploid and homoploidhybrids will be positive for chlorophyll pigmentation in corollatissue if at least one progenitor has chlorophyll present in itspetals (due to evidence of the dominance of chlorophyll pig-mentation) (3) increased cell size potentially associated withpolyploidy affects the concentration of pigments and in turncolour intensity and (4) floral colour evolution is constrainedby phylogeny
1118 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
N tabacum 095-55
N tabacum 51789
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N sylvestris N tomentosiformis
N tabacum lt0middot2 myo
times
Natural Synthetic
A
timesN sylvestris N otophora TH32
TH32 synthetic polyploid
Synthetic
B
timesN rustica
var asiaticaN rustica
var pavoniisynthetic
UxPsynthetic PUE1 F1
synthetic N rusticaPUE1-R10
S0
synthetic N rustica
PUE1-R1 S1
N paniculata N undulata
N rustica lt0middot2 myo
Natural
Synthetic
C
timesN undulata N wigandioides N arentsii
N arentsii lt0middot2 myo
Natural
D
Natural Synthetic
timesN clevelandii N quadrivalvis
904750042N quadrivalvis
TW18N obtusifolia
lsquoBaldwinrsquo
N attenuata
lsquoBaldwinrsquo
N times obtusiata line 1
N times obtusiata line 2
N times obtusiata line 5
Section Polydicliae ~1 myoE
timesN repanda N nesophila N stocktonii N nudicaulisN sylvestris N obtusifolia
TW143
Section Repandae~4middot5 myoF
Natural
NoctifloraendashPetunioides homoploid hybrids
N noctiflora N petunioides N acuminata N attenuata N miersii N pauciflora N linearis TW77
N glauca 51725
N glauca 51751
times
Section PetunioidesSection Noctiflorae
H
N glutinosa
N otophora N setchellii N tomentosiformis N undulata N wigandioides N glutinosa
times
Section Tomentosae Section Undulatae
I
Section Suaveolentes ~10 myo
N noctiflora N petunioides N acuminata N attenuata N miersii N pauciflora
times
G
Section PetunioidesSection Noctiflorae
times
N sylvestris
N benthamiana
N forsteri
N gossei
N occidentalis
var hesperis
N megalosiphonN suaveolens
FIG 1 Floral colour as perceived by humans of polyploid and homoploid hybrid Nicotiana and their diploid progenitors Polyploid ages were estimated using a mo-lecular clock calibrated with the geological age of volcanic islands with endemic Nicotiana species (Clarkson et al 2005) Absolute dates (millions of years myo)estimated by the clock should be treated with caution however relative ages of different polyploid sections should reflect the true sequence of polyploidizationevents (A) Natural and synthetic polyploids of N tabacum (B) Synthetic polyploid TH32 (C) Natural and synthetic N rustica polyploids Synthetic hybrids includea homoploid from a reciprocal cross and a polyploid series (F1 homoploid and S0 and S1 polyploids) of the same parentage as natural N rustica (D) Nicotiana are-ntsii (E) Natural polyploids of section Polydicliae Synthetic N obtusiata polyploid lines were made from a cross between the N obtusifolia and N attenuata ac-cessions studied here (F) Section Repandae (G) Section Suaveolentes contains 26 polyploid species (six included in this study) Biogeographical analyses suggestthat section Suaveolentes originated15 million years ago (mya) before the aridification of Australia (Ladiges et al 2011) and this seems to be relatively congru-ent with the molecular clock results which place its origin at 10mya (H) Homoploid hybrids N glauca and N linearis (I) Homoploid hybrid N glutinosa
Photographs are scaled to the same size
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1119
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
MATERIALS AND METHODS
Petal cell area measurements
To assess whether an increase in ploidy results in larger petalcells cell area was measured from a subset of polyploids andtheir progenitors For Nicotiana sylvestris A04750326Nicotiana rustica var asiatica Nicotiana rustica var pavoniiNicotiana paniculata Nicotiana undulata and Nicotiana nudi-caulis mature flowers were taken from plants and the adaxialpetal surface was imprinted in Elite HD vinylpolysiloxane im-pression material (dental wax supplied by ZhermackHarrogate UK) The wax was left to set and then used as amould for making epoxy petal casts Devcon high-strength ep-oxy was mixed according to the manufacturerrsquos instructionspoured into the mould and allowed to set for 12 h The epoxyrelief was removed and coated with gold using a QuorumK756X sputter coater The samples were then imaged using aFEI Philips XL30 FEGSEM scanning electron microscope ForNicotiana obtusifolia var obtusifolia TW143 Nicotianarepanda and Nicotiana stocktonii only fixed material wasavailable whole mature flowers were fixed in formalinndashaceticacidndashalcohol (FAA) (60 ethanol 6 formaldehyde 5 acetic acid) for 72h before being transferred to a 70 ethanol(EtOH) wash for 24 h The samples were then dehydratedthrough an ethanol series of 2 h each in 70 80 and 90 andtwo washes in 100 EtOH The samples were dissected andthen dried in an Autosamdri 815B critical point dryer Thesesamples were sputter-coated and imaged as described aboveFor all samples images were taken mid-petal from an angleperpendicular to the surface to minimize parallax error Cellsize measurements were carried out in ImageJ (httpimagejnihgovij) The circumference of the cell base was drawn free-hand and area was calculated for 100ndash150 cells until the cu-mulative mean stabilized One-way ANOVA and Tukeyrsquoshonest significance tests were performed in RStudio version098490 (httpwwwrstudioorg) to compare cell area of poly-ploids with those of their progenitors repeating the tests foreach polyploid section
Spectral reflectance measurements
Spectral reflectance measurements were recorded for 60Nicotiana accessions (41 taxa Supplementary Data Table S1)three flowers from different plants where possible were usedfor each accession Reflectance spectra from three Nicotianaotophora accessions were pooled because the spectra weresimilarSpectral reflectance of flowers at anthesis was measured
from 300 to 700 nm using an Avantes AvaSpec-2048 spectro-photometer with an Avantes AvaLight-DHS light source andcalibrated with a barium sulphate white standard from lab-sphereVR Nicotiana mutabilis was also measured later as flow-ers change from white to pink when mature pink flowers areless likely to have a nectar reward but add to the attraction ofthe overall floral display and therefore are still relevant to pol-linators (R Kaczorowski University of Haifa Israel perscomm) Reflectance spectra express the proportion of lightreflected by the flower at any given wavelength Spectra werevisualized and exported in increments of 1 nm using the
program AvaSoft version 703 Full (Avantes BV Eerbeek TheNetherlands) and imported into ExcelSpectra for each accession or colour morph were averaged
and then smoothed three times using a rolling average over9 nm Spectra for all accessions were submitted to the FloralReflectance Database (FReD wwwreflectancecouk Arnoldet al 2010)Some spectra had a spike at 656 nm which corresponded
to a narrow peak in the light source spectrum suggesting thatthe spectra were saturated at 656 nm however smoothingserved to neutralize this spike Several spectra (Nicotiana are-ntsii N mutabilis Nicotiana suaveolens and Nicotiana wigan-dioides) included an anomalous reflectance minimum from 475to 500 nm which could not be explained by the light sourcespectrum Remeasured spectra of N arentsii N suaveolens andN wigandioides lacked this minimum but further material ofN mutabilis was unavailable so these spectra were includeddespite the anomalies
Calculation of colour loci in the bee colour hexagon
A reflectance spectrum can be represented as a single pointin the bee colour hexagon space (a graphical representation of abeersquos colour visual experience) based on the relative excitationof UV- blue- and green-sensitive photoreceptor types (Chittka1992) Vertices of this hexagon represent theoretical stateswhere one or two photoreceptor types are at maximal excitationwhereas at least one receptor type is at zero excitation (eg thetop vertex of the hexagon corresponds to maximal blue receptorexcitation and zero signal from UV and green receptorswhereas the top right vertex corresponds to maximal signal inboth blue and green receptors but no signal in the UV receptorand so forth see Supplementary Data Fig S1) The centre ororigin of the hexagon is achromatic Hue corresponds to angu-lar position around the origin whereas spectral purity or satura-tion increases with distance from the originBee colour hexagon coordinates were calculated for all
Nicotiana spectra Illumination was assumed to be sunlight(D65 Wyszecki and Stiles 1982) the background was repre-sented by an average leaf spectrum (Gumbert et al 1999)Honeybee photoreceptor spectral sensitivity functions wereused to determine bee colour hexagon coordinates these aresimilar to bumblebee and hawkmoth photoreceptor sensitivityfunctions so the bee colour hexagon can be used to approxi-mate the colour vision of these insects as well (Menzel et al1986 Peitsch et al 1992 Briscoe and Chittka 2001 Kelberet al 2003 and references therein Skorupski et al 2007) Theequations used to determine colour hexagon coordinates are asfollows where EG EB and EUV represent the excitation of thegreen blue and UV bee photoreceptors respectively elicitedby a spectrum (Chittka 1992)
x frac14ffiffiffi
3p
=2ethEG EUVTHORNy frac14 EB 05ethEUV thorn EGTHORN
Because the colour loci of Nicotiana flowers were mostlyclose to the centre of the colour space all colour hexagon dis-plays presented are scaled so that only the central 40 is
1120 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
shown the outline is therefore drawn as a dashed line This re-sults in a clearer spread of the colour loci to facilitate visual in-spection For a diagram explaining the colour hexagon seeSupplementary Data Fig S1
Calculation of colour loci in hummingbird colour space
For tetrachromatic hummingbirds we chose to model flowercolours in a 3-D colour opponent space because n ndash 1 colouropponent dimensions are necessary to code the informationfrom n colour receptors (Chittka 1996) The hummingbirdcolour space can be displayed as a rhombic dodecahedron with14 vertices (Restrepo 2013) Like the bee colour hexagonvertices of the space represent states where one two or threephotoreceptor types are at maximal excitation and at least onereceptor type is at zero excitation
Hummingbird colour space coordinates were calculated forall Nicotiana spectra Illumination was again assumed to besunlight (D65 Wyszecki and Stiles 1982) and the backgroundan average leaf spectrum (Gumbert et al 1999) as was usedfor the bee colour hexagon Photoreceptor spectral sensitivityfunctions from the green-backed firecrown hummingbird(Sephanoides sephanoides Herrera et al 2008) were used todetermine hummingbird colour space coordinates usingthe following equations (Restrepo 2013) where ER EGEB and EUV represent excitation of red green blue andUV hummingbird photoreceptors respectively elicited by aspectrum
x frac14ffiffiffiffiffiffiffiffi
3=4p
EB ffiffiffiffiffiffiffiffiffiffi
1=12p
ethEUV thorn EG thorn ERTHORN
y frac14ffiffiffiffiffiffiffiffi
2=3p
EG ffiffiffiffiffiffiffiffi
1=6p
ethEUV thorn ERTHORN
z frac14ffiffiffiffiffiffiffiffi
1=2p
ethEUV ERTHORN
RStudio was used to make 3D plots of the hummingbird col-our space and ImageJ version 148 (httpimagejnihgovij)was used to create an animation of the Nicotiana flower loci inthe hummingbird colour space Again Nicotiana flower colourloci are close to the origin in the hummingbird colour space sothe graphs presented display only the central portion (either 25or 50 ) of the colour space for clarity To further facilitateinterpretation of these graphs vertices representing individualexcitation of the red green blue and UV photoreceptor typesas well as their excitation vectors from the origin are shownin red green blue and black respectively Other vertices(representing excitation of two or three photoreceptor types) areshown in grey
Cluster analyses
Cluster analyses were used to group spectra based on spectralshape (corresponding to pigmentation) and their position inboth bee and hummingbird colour spaces For spectral colourcategories spectra were normalized to the same integral underthe curve in order to compare combinations of pigments notthe concentration of pigments A distance matrix was calculatedfrom the normalized spectral data in R version 302 (http
wwwR-projectorg) using the dist() function For the bee andhummingbird colour categories the input data were the (x y) or(x y z) coordinates of the spectra in the bee and hummingbirdcolour spaces respectivelyData were first imported into R The function hclust() was
used to perform agglomerative hierarchical clustering based onthe average pairwise distances between groups With this algo-rithm the observed points which are initially all deemed to bedistinct are iteratively assigned to groups until eventually allpoints belong to the same group At each step the average dis-tance between all groups (ie the mean distance from all pointsin group A to all points in group B if either one of these is asingle point then no averaging is needed) is calculated and thetwo groups with the smallest average distance are merged Theorder in which groups are merged can be used to construct adendrogram showing the spatial relationship between all datapoints We can also look at the distribution of merge distancesat each step in the algorithm and use this distribution to esti-mate how many groups are present in the data Points at whichthere is a steep increase in the average between-group distance(lsquoelbowrsquo points) highlight the spatial scale(s) at which there isclustering present in the data By using one of these elbowpoints as a cutoff in the algorithm we can arrive at a distancegrouping that captures the spatial clustering It should be notedthat the determination of where to draw the threshold in a clus-ter analysis is arbitrary but the use of one of these elbow pointsdoes yield meaningful clusters The determination of the spe-cific point from the elbow region to be used to define clusterswas further informed by visual inspection of reflectance spec-tra as well as distributions of colour loci in the perceptual col-our spaces It should be noted that the dendrograms relate tosimilarities in spectral reflectance as well as colour relation-ships perceived by bees and hummingbirds they do not showphylogenetic relationships
Ancestral state reconstruction
To examine the evolution of colour within a phylogeneticcontext ancestral state reconstructions were performed on treesinferred from plastid sequence data Only species for which flo-ral character data are available were included in these analysesBecause polyploid and homoploid hybrids originate via reticu-late evolutionary processes and therefore lack a history oftree-like evolution ancestral characters were reconstructed us-ing only non-hybrid diploid species The states observed inhybrid species were then compared with the ancestral state re-constructions Since sections Repandae and Suaveolentes havediversified to form several species following polyploidizationcharacters were reconstructed for these sections separately toexamine colour evolution subsequent to their origin Fornon-hybrid diploid species individual gene trees yieldsome conflicting topologies nevertheless key nodes for thepurposes of interpreting character evolution in hybrids arerecovered in multiple gene trees and are supported bysupernetwork analyses (Kelly et al 2010) Therefore plastiddata from previously published studies are suitable for theseanalysesPreviously published sequences (Clarkson et al 2004) from
four plastid regions (matK ndhF trnL-F and trnS-G) were
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1121
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
aligned separately using PRANKthornF (Loytynoja and Goldman2008) and then concatenated to create a combined plastid data-set before further optimization by eye in Mesquite version 274(Maddison and Maddison 2008) For Nicotiana attenuata weused GenBank accessions AB040009 and AY098697 for thematK and trnL-F regions respectively (due to likely misidenti-fication of N attenuata material used in Clarkson et al 2004see Clarkson et al 2010) the other two regions were scored asmissing data Phylogenetic reconstruction by Bayesian infer-ence was performed as described in Kelly et al (2013) with theexception that BayesTrees v13 (wwwevolutionreadingacukBayesTreeshtml) was used to construct 95 majority ruleconsensus trees For sections Repandae and Suaveolentes se-quences representing their putative maternal progenitors wereincluded during Bayesian inference to allow rooting of trees butwere pruned from the trees prior to ancestral statereconstructionAncestral states for spectral reflectance colour categories and
presenceabsence of chlorophyll in petals (data inSupplementary Data Table S2) were reconstructed using theparsimony reconstruction method in Mesquite version 274 un-der the unordered states assumption To account for topologicaluncertainty character states were reconstructed over all 36 000post-burn-in trees using the Trace Character Over Trees optionand summarized on the 95 majority rule consensus tree fromthe Bayesian analysis Ancestral states were not calculated forbee or hummingbird colour categories because these are per-ceptual systems and the same colour category can result fromdifferent combinations of pigments (eg both human pink andhuman white flowers which are positive and negative respec-tively for anthocyanin pigmentation are both classified as beebluendashgreen) thus a single colour category does not necessarilyhave a shared evolutionary history
Estimating expected polyploid and homoploid hybrid floralcolour
Polyploid and homoploid hybrid floral colours for each ac-cession were compared with those of their diploid progenitorsfor the spectral bee and hummingbird colour categories definedby cluster analyses Floral colour was classified as lsquoexpectedrsquo ifit fell in the colour category of at least one progenitor or lsquounex-pectedrsquo if it was different from both progenitors Polyploid andhomoploid hybrids were also compared with their diploid pro-genitors for the presence or absence of chlorophyll in corollatissue Chlorophyll absorbs at 675 nm in vivo (Haardt andMaske 1987) therefore the presence of chlorophyll can be in-ferred from reflectance spectra if there is a reflectance mini-mum at 675 nm The presence of chlorophyll in petal tissueappears to be dominant (Brieger 1935 Ohmiya et al 2014)Thus hybrids were classified as expected if they showed chlo-rophyll pigmentation and as unexpected if they did not becauseat least one progenitor possessed chlorophyll in all diploid pro-genitor combinations For natural homoploid hybrids and poly-ploid section Suaveolentes where progenitors can only bedefined to Nicotiana section level comparisons were madewith reconstructed ancestral characters thus only spectral col-our categories and the presenceabsence of chlorophyll wereexamined
Phylogenetic signal in floral traits
In order to statistically test for phylogenetic signal in the phe-notypic trait data (spectral reflectance bee and hummingbirdcolour perception) we used Mantel tests to examine the corre-lation between phylogenetic distance and each of the respectivecontinuous multidimensional traits (eg Cubo et al 2005Muchhala et al 2014) Analyses were restricted to diploid spe-cies excluding homoploid and polyploid hybrids Trees wereedited in Newick format to include additional tips with zerobranch lengths for taxa that are multiple in the trait datasetseither due to colour polymorphism (N otophora) or multipleaccessions (N sylvestris and N obtusifolia var obtusifolia)Statistical analyses were performed in R version 310
Phenotypic distance matrices were first calculated for the threetrait datasets using Euclidean distance and phylogenetic dis-tance matrices were calculated (1) as genetic distance from theplastid alignment and (2) for each of 36 000 post-burn-inBayesian trees using copheneticphylo() part of the APE pack-age version 31-2 (Paradis et al 2004) The second Bayesianset of tests was performed in order to account for evolutionaryprocesses such as saturation and to estimate how phylogeneticuncertainty affects the correlation Mantel tests were performedusing Pearsonrsquos product-moment correlation coefficient with10 000 permutations of each distance matrix to test for signifi-cance the mean P value and its standard deviation were calcu-lated for each set of 36 000 Mantel tests from the Bayesiantrees along with the percentage of trees that gave significantcorrelations The function mantel() from the vegan packagewas used (Oksanen et al 2013)
RESULTS
Petal cell area
Petal cell area was measured to determine whether an increasein ploidy results in larger floral cells Polyploid petal cell areawas significantly larger than in both progenitors in N rustica(ANOVA Ffrac14 371 dffrac14 3 Plt 2 10ndash16) accessions but wasintermediate between progenitors in section Repandae poly-ploids (ANOVA Ffrac14 2492 dffrac14 4 Plt 2 10ndash16 Fig 2)Accessions that were significantly different in cell area (withinpolyploid sections and their progenitors) are represented bydifferent letters above the bars in Fig 2 results from Tukeyrsquoshonest significance tests can be found in Supplementary DataTable S3
Cluster analyses
Nicotiana reflectance spectra were grouped into categoriesbased on spectral shape and position in the bee and humming-bird colour spaces using cluster analyses Bees and humming-birds have different photoreceptor sensitivities and we expectour cluster analyses to reflect these differences in sensoryequipment The analysis based on spectral shape yielded eightcolour categories which roughly corresponded to flowers per-ceived by human observers as magenta red pink UVndashwhitewhite yellow green and dark green (Fig 3) Nicotiana spectraare displayed by spectral colour category in Fig 4A BSupplementary Data Fig S2 The bee colour hexagon clustering
1122 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
resulted in 11 colour categories which fell into the followingareas of bee colour space saturated green UVndashblue high UVUVndashgreen green light green bluendashgreen dark green saturatedUVndashblue saturated UVndashgreen and blue (the last four categorieswere each represented by only a single accessionSupplementary Data Fig S3A) These groups are shown in thebee colour hexagon (Fig 4C) The hummingbird colour spacecluster analysis also produced 11 colour categories saturatedgreen green UVndashwhite UVndashgreen pink white UVndashpinkdark green light pink red and saturated UVndashpink (again thelast four categories include only a single accessionSupplementary Data Fig S3B) These groups are shown in thehummingbird colour space (Fig 4D) and the same graph isprovided as an animation to better display the 3D nature of thecolour space (Supplementary Data Video)
Evolution of spectral reflectance in polyploids and homoploidhybrids
To assess the evolution of polyploid floral colour polyploidand homoploid hybrid accessions were compared with those oftheir progenitors in spectral bee and hummingbird colour cate-gories as well as in the presenceabsence of chlorophyll Thediploid progenitors and approximate age of polyploids andhomoploid hybrids are found in Fig 1 and Supplementary DataTable S4 and the observed and expected floral colours of
polyploid and homoploid hybrids are found in Table 1 Mostpolyploid and homoploid hybrids were similar to at least oneprogenitor in spectral bee and hummingbird colour categoriesbut some fell into unexpected colour categories (Table 1Fig 5 Supplementary Data Fig S4 Supplementary Data FigS5) Over half of the polyploids unexpectedly lacked chloro-phyll (Table 1)
Evolution of colour characters in a phylogenetic context
Reconstructed character states are shown for spectral reflec-tance colour categories (Fig 6) and the presenceabsence ofchlorophyll in petals (Supplementary Data Fig S6) Bee andhummingbird colour categories are also shown for extant spe-cies on the plastid tree (Fig 6) Although the deepest nodeswere largely equivocal evolution of spectral reflectance colourin Nicotiana seemed to be dynamic (Fig 6) Green flowerslikely have three independent origins (1) in sectionsPaniculatae and Undulatae (2) in N langsdorffii and (3) inthe homoploid hybrid N glauca UVndashwhite flowers also seemto have arisen three times independently (1) in sectionTrigonophyllae (2) in N pauciflora and (3) in the homoploidhybrid N linearis Most polyploid and homoploid hybrid spe-cies exhibit a floral colour present in at least one of their pro-genitors However N tabacum 095-55 is red and N glauca isyellow and green unlike their progenitors UVndashwhite flowersseem to have evolved de novo in N linearis UVndashwhite flowersare also found in one of its progenitor sections (in N pauci-flora) but ancestral reconstructions indicate that the floral col-our was most likely white at the ancestral nodes within thesection (Fig 6) This suggests that the evolution of UVndashwhiteflowers in N pauciflora has occurred subsequent to the forma-tion of N linearis and that the two events are likely indepen-dent It is unclear whether UVndashwhite flowers also evolved denovo in N nudicaulis because the ancestral node of sectionRepandae is equivocal The presence of chlorophyll as inferredby light absorption at 675 nm (Haardt and Maske 1987) inNicotiana flowers is ancestral and has been lost three times inN sylvestris N noctiflora and the most recent common ances-tor of N acuminata and N pauciflora (Supplementary DataFig S6)Results from Mantel tests of phylogenetic signal for
Nicotiana floral traits for both genetic distance and the 36 000post-burn-in Bayesian trees are shown in Table 2 All floraltraits were significantly correlated with phylogenetic relation-ships for the Bayesian trees at a significance level of Plt 005Only spectral reflectance was significant for the genetic dis-tance tests whereas bee and hummingbird colour perceptionwere just above the Plt 005 threshold For the Bayesian trees901 662 and 932 of trees were significantly correlatedwith the spectral reflectance bee and hummingbird colour per-ception datasets respectively These results suggest that thesefloral traits are weakly constrained by phylogeny
DISCUSSION
Nicotiana is remarkable in its range of spectral reflectanceflower colours (white UVndashwhite pink magenta red yellowgreen and dark green Fig 3) and in the variety of pollinators
0
200
400
600
800
1000
1200C
ell
are
a (
microm
2)
N rustica
N paniculata
N undulata
N sylvestris
N obtusifolia
N nudicaulis
N repanda
N stocktonii
N rustica var asiatica
N rustica var pavonii
Repandae
a
a
b
b
cc
c c
d
Maternal progenitor
Paternal progenitor
Polyploids
Polyploids
Polyploids
FIG 2 Petal cell area from polyploids and their progenitors Within each poly-ploid group bars with different letters represent significantly different mean cell
areas
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1123
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
that visit the flowers (moth bird bee bat Knapp 2010) Theperception of these spectral colours also changes with visualsystem (bee or hummingbird) Here we describe a complex dy-namic in the evolution of floral colour in Nicotiana Spectral re-flectance and bee and hummingbird colour perception arecorrelated with phylogeny but multiple independent origins ofvarious combinations of pigmentation suggest that the evolutionof floral colour is not entirely phylogenetically constrained
Petal cell size evolution in polyploids
Cell size is expected to increase following polyploidizationdue to the increase in genome size (Beaulieu et al 2008) Thesignificantly larger petal cells of N rustica (lt02 million yearsold myo) and the intermediate cell size of section Repandae(45myo) suggest that polyploids may revert to a diploid-like cell size over time similar to the genome downsizing ob-served in polyploids (Leitch and Bennett 2004) Howeverpetal cell size differences within section Repandae do not seemto be linked to genome size N nudicaulis and N repanda sharesimilar cell sizes but have substantially different genome sizes(Leitch et al 2008) The concentration of pigment in floralcells is controlled by both the amount of pigment present andthe cell size Nicotiana rustica has significantly larger petalcells than its progenitors (close to the sum Fig 2) and displaysan intermediate brightness (the area under the reflectance curvea proxy for pigment concentration) between its progenitors
(Supplementary Data Fig S4D) This is expected if the poly-ploid inherits the sum of both cell size and amount of pigmentpresent from its progenitors
Polyploid divergence in floral colour
Many younger polyploids (lt02myo) display unexpectedfloral colours considering those of their diploid progenitorsNone of the natural and synthetic N tabacum accessions pos-sess chlorophyll which is unexpected given its presence in atleast one progenitor species Nicotiana tabacum 095-55 alsohas unexpected spectral bee and hummingbird colour giventhe colour categories of the progenitor species Similarly syn-thetic N tabacum QM has unexpected colours in bee and hum-mingbird perception Because this accession is synthetic theparents are known and thus its unexpected phenotype can beclassified as transgressive or outside the range of its progeni-tors due solely to polyploidy and hybridization Most N rusticaaccessions have unexpected bee colour (four of these aresynthetic and are therefore also transgressive) and N arentsiihas unexpected hummingbird colour (Table 1 Fig 6) Despitethe divergence of floral spectra associated with polyploidybehavioural studies are needed to determine whether the beeand hummingbird colour categories delineated here actuallyelicit different responses in pollinatorsMost older polyploids (1ndash10myo) are similar in floral col-
our category to at least one of their progenitors N clevelandii
synth
etic
N ru
stic
a P
UE
1-R
1 S
1
N ru
stic
a v
ar a
sia
tica
synth
etic
N ru
stic
a P
UE
1-R
10 S
0
synth
etic
PU
E1 F
1synth
etic
UxP
N b
enavid
esii
N ru
stic
a v
ar p
avonii
N p
anic
ula
taN
langsdorffii
N g
lauca 5
1751 g
reen
N u
ndula
taN
raim
ondii
N k
nig
htia
na
N g
lauca 5
1751 y
ello
w
N g
lauca 5
1725
N p
lum
bagin
ifolia
N g
ossei
N q
uadriv
alv
is T
W18
N o
ccid
enta
lis s
ubsp h
esperis
N
megalo
sip
hon
N s
uaveole
ns
N n
esophila
N s
tockto
nii
N q
uadriv
alv
is 9
04750042
N a
ttenuata
N re
panda
N s
ylv
estris
A04750326
N
benth
am
iana
N m
uta
bilis
CP
G3 w
hite
N
muta
bilis
CP
G12456 w
hite
N
wig
andio
ides
N a
cum
inata
N fo
rste
riN
cle
vela
ndii
N n
octiflo
raN
x o
btu
sia
ta lin
e 2
N
x o
btu
sia
ta lin
e 1
N
mie
rsii
N p
etu
nio
ides
N x
obtu
sia
ta lin
e 5
N
sylv
estris
6898
N
oto
phora
white
N
are
nts
iiT
H3
2N
tabacum
lsquoChulu
manirsquo
N o
btu
sifo
lia v
ar p
alm
eri
N o
btu
sifo
lia v
ar o
btu
sifo
lia lsquoB
ald
win
rsquoN
linearis
TW
77
N n
udic
aulis
N o
btu
sifo
lia v
ar o
btu
sifo
lia T
W143
N lin
earis
964750099
N
pauciflo
raN
tom
ento
sifo
rmis
N o
tophora
pin
k
N g
lutin
osa
synth
etic
N ta
bacum
TH
37
N
tabacum
51789
synth
etic
N ta
bacum
QM
N
tabacum
095-5
5
N s
etc
hellii
N m
uta
bilis
CP
G3 p
ink
N m
uta
bilis
CP
G12456 p
ink
10
8
6
4
2
0
Dis
tan
ce
Spectral categories
MagentaRedPinkUV-white
WhiteYellowDark greenGreen
Dark greenSaturated greenGreenUV-whiteUV-green
PinkLight pinkWhiteUV-pinkRedSaturated UV-pink
SpecBeeHum
SpectralDark greenSaturated greenSaturated UV-blueUV-blueSaturated UV-greenHigh UV
UV-greenGreenLight greenBlueBlue-green
Bee Hummingbird
FIG 3 Dendrograms based on distance cluster analyses for spectral reflectance Coloured circles on the dendrogram represent distinct colour categories as determinedby the chosen threshold (dashed line) Dendrograms were similarly constructed for bee and hummingbird colour (see Supplementary Data Fig S3) The lines of col-oured circles at the tips of the dendrograms signify the category each taxon is assigned to in spectral bee and hummingbird colour as labelled (Spec Bee and Hum)for comparison between spectral categories and those of different visual systems Diploids polyploids and homoploids are denoted by black blue and orange text
respectively
1124 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
0
0middot2
0middot4
0middot6
0middot8
1middot0
Reflecta
nce
Wavelength (nm)
0
0middot2
0middot4
0middot6
0middot8
1middot0
300 350 400 450 500 550 600 650 700 300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
N benavidesii
N glauca 51751 g
N langsdorffii
N paniculata
N raimondii
N rustica var asiatica
N rustica var pavonii
synthetic UxP
synthetic F1
syn N rustica S0
syn N rustica S1
N undulata
N glutinosa
N otophora p
N tabacum 51789
syn N tabacum QM
syn N tabacum TH37
N tomentosiformis
A B
G 50
R 50
B 50
UV 50
Hummingbird colour space
White
0middot40middot4
ndash0middot40middot4
0middot4
0middot4
y
z
UV-white
Pink
Synthetic N tabacum QM
UV-pink
N tabacum 095-55
N setchellii
Green
N langsdorffii
Saturated green
UV-green
C DBee colour hexagon
Blue-green
Light green
Green
Dark green
Sat green
UV-green
Sat UV-green
B
B-G
G
UV-G
UV
UV-B
1 acum2 aren3 atten4 bena5 benth6 clev7 forst8 glau259 glau51y10 glau51g11 glut12 goss13 knight14 langs15 lin964716 linTW7717 mega18 mier19 mut1w20 mut1p21 mut3w22mut3p23 neso24 noct25 nudi26 xobtus127 xobtus228 xobtus529 obtusB30 obtusTW31 obtuspalm
32 occhesp33 otoph w34 otoph p35 pani36 pauc37 petun38 plumba39 quad904740 quadTW1841 raim42 repa43 rustasi44 rustpav45 syn UxP46 syn F147 syn rust S048 syn rust S149 setch50 stock51 suav52 sylv689853 sylvA04754 tab0955555 tab5178956 tabchulu57 syntabQM58 syntabTH3759 tomtform60 undu61 wigan62TH32
High UV
UV-blue
Sat UV-blue
Blue
FIG 4 (A B) Nicotiana reflectance spectra from 300 to 700nm which roughly correspond to colours perceived by human observers as pink (A) and green (B) SeeSupplementary Data Fig S2 for other spectral colour categories Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploidhybrid taxa Abbreviations p pink syn synthetic g green (C) Colour hexagon displaying the distribution of Nicotiana colour loci in bee colour space The hexa-gon has been scaled so that vertices represent 40 excitation of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-GUVndashgreen Bee colour categories are delineated by coloured ovals sat saturated Nicotiana species abbreviations are as follows acum acuminata aren arentsiiatten attenuata benavid benavidesii benth benthamiana clev clevelandii forst forsteri glau25 glauca 51725 glau51y glauca 51751 yellow glau51g glauca51751 green glut glutinosa goss gossei knight knightiana langs langsdorffii lin9647 linearis 964750099 linTW77 linearis TW77 mega megalosiphon miermiersii mut1w mutabilis CPG12456 white mut1p mutabilis CPG12456 pink mut3w mutabilis CPG3 white mut3p mutabilis CPG3 pink neso nesophila noctnoctiflora nudi nudicaulis obtus1 obtusiata line 1 obtus2 obtusiata line 2 obtus5 obtusiata line 5 obtusB obtusifolia var obtusifolia lsquoBaldwinrsquoobtusTW obtusifolia var obtusifolia TW143 obtuspalm obtusifolia var palmeri occhesp occidentalis subsp hesperis otoph w otophora white otoph p otophorapink pani paniculata pauc pauciflora petun petunioides plumba plumbaginifolia quad9047 quadrivalvis 904750042 quadTW18 quadrivalvis TW18 raimraimondii repa repanda rustasi rustica var asiatica rustpav rustica var pavonii syn UP synthetic UP syn F1 synthetic PUE1 F1 synrust S0 synthetic rus-tica PUE1-R10 S0 synrust S1 synthetic rustica PUE1-R1 S1 setch setchellii stock stocktonii suav suaveolens sylv6898 sylvestris 6898 sylvA047 sylvestrisA04750326 tab09555 tabacum 095-55 tab51789 tabacum 51789 tabchulu tabacum lsquoChulumanirsquo syntabQM synthetic tabacum QM syntabTH37 synthetictabacum TH37 tomtform tomentosiformis undu undulata wigan wigandioides TH32 TH32 synthetic N sylvestrisN otophora polyploid (D) Distribution ofNicotiana spectral loci in hummingbird colour space Vertices of the hummingbird colour space represent 50 excitation of the photoreceptors single photoreceptortype vertices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows repre-sent the vectors of these photoreceptors from the origin of the hummingbird colour space Nicotiana loci are coloured according to hummingbird colour categories(Supplementary Data Fig S3B) but are labelled with the accession name if the category includes only one taxon See Supplementary Data Video for an animation
of this graph
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1125
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
is the exception because it falls into an unexpected humming-bird colour category given the progenitor species (Fig 6)Section Repandae polyploids seem to have evolved to be eitherlike their maternal (Nicotiana nesophila N repanda andN stocktonii) or paternal (N nudicaulis) progenitor (Fig 6)The maternal progenitor N sylvestris is no longer sympatricwith any of the section Repandae polyploids therefore Nnesophila N repanda and N stocktonii can occupy the samepollination niche as their maternal progenitor without competi-tion Similarly section Suaveolentes is native to Australasiaexcept for one species in Namibia Africa and is not sympatricwith its progenitor sections in South America (Goodspeed1954) these polyploids and their diploid progenitors displaysimilar floral colours except N pauciflora which evolvedspectrally UVndashwhite flowers after the formation of sectionSuaveolentes (Fig 6) It is possible that there is less competitionfor pollinators and therefore reduced selective pressure to-wards floral colour diversification when polyploid species arenot sympatric with their diploid progenitors as is seen in
Iochrominae (Solanaceae) which have a broader range of floralcolours when species are sympatric (Muchhala et al 2014)However floral colour evolution can also be driven by geneticdrift or selection on pleiotropic effects of floral genes (Chittkaet al 2001 Rausher 2008) Furthermore anthocyanins andflavonoids are important for UV protection can defend againstfungi act as signal molecules and play a role in male fertility insome species in addition to their roles in signalling to pollina-tors (Shirley 1996)Over half of polyploids have an unexpected inheritance pat-
tern for chlorophyll if it is assumed that the presence of chloro-phyll is a dominant character Those polyploids that deviatefrom expectation span an age range of synthetic to 10myoand always lack chlorophyll in their petals which is in linewith the direction of the shifts observed in the divergence ofdiploid species (Supplementary Data Fig S6) In carnation thedifference in chlorophyll concentration between white andgreen flowers is likely caused by downregulation of chlorophyllbiosynthesis genes in white flowers genes involved in
TABLE 1 Polyploid and homoploid hybrid observed and expected floral colours
Species Spectral Bee Hummingbird Chlorophyll
Observed Expected Observed Expected Observed Expected Observed Expected
N tabacum 095-55 R W P UV-B B-G sUV-P W P N CN tabacum 51789 P W P B-G B-G P W P N CN tabacum lsquoChulumanirsquo W W P B-G B-G P W P N Csyn N tabacum QM P W P B B-G LP W P N Csyn N tabacum TH37 P W P B-G B-G P W P N CTH32 P W P B-G B-G P W P N CN rustica var asiatica G G LG G B-G G G W C CN rustica var pavonii G G G G B-G G G W C Csyn UP (homoploid) G G UV-G G B-G UV-G G W C Csyn F1 (homoploid) G G LG G B-G G G W C Csyn N rustica S0 G G LG G B-G G G W C Csyn N rustica S1 G G LG G B-G G G W C CN arentsii W G W B-G B-G P W C CN clevelandii W UV-W W B-G UV B-G P UV-W W N CN quadrivalvis TW18 W UV-W W B-G UV B-G W UV-W W C CN quadrivalvis 9047 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 1 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 2 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 5 W UV-W W B-G UV B-G W UV-W W C CN repanda W W UV-W B-G B-G UV W W UV-W N CN nesophila W W UV-W B-G B-G UV W W UV-W N CN stocktonii W W UV-W B-G B-G UV W W UV-W C CN nudicaulis UV-W W UV-W UV B-G UV UV-W W UV-W C CN benthamiana W W ndash ndash ndash ndash N CN forsteri W W ndash ndash ndash ndash C CN gossei W W ndash ndash ndash ndash N CN megalosiphon W W ndash ndash ndash ndash N CN occidentalis W W ndash ndash ndash ndash N CN suaveolens W W ndash ndash ndash ndash N C
N glauca 51725 Y W ndash ndash ndash ndash C CN glauca 51751 YG W ndash ndash ndash ndash C CN linearis TW77 UV-W W ndash ndash ndash ndash C CN linearis 9647 UV-W W ndash ndash ndash ndash C CN glutinosa P P W G ndash ndash ndash ndash C C
The top block consists of polyploid accessions and the bottom block includes homoploid hybridsItalic denotes an unexpected phenotype given the colour categories of the progenitorsProgenitor bee and hummingbird colour categories are unknown for section Suaveolentes and natural homoploid hybrids (see text)R red W white P pink G green UV-W UVndashwhite Y yellow UV-B UVndashblue B-G bluendashgreen B blue LG light green UV high UV UV-G
UVndashgreen sUV-P saturated UVndashpink LP light pink N no chlorophyll C chlorophyll syn synthetic 9047 represents 904750042 9647 represents964750099
1126 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
The effect of polyploidy and hybridization on the evolution of floral colourin Nicotiana (Solanaceae)
Elizabeth W McCarthy1ndash3y Sarah E J Arnold1z Lars Chittka1 Steven C Le Comber1 Robert Verity1xSteven Dodsworth13 Sandra Knapp2 Laura J Kelly13 Mark W Chase3 Ian T Baldwin4 Ales Kovarık5
Corinne Mhiri6 Lin Taylor7 and Andrew R Leitch11School of Biological and Chemical Sciences Queen Mary University of London Mile End Road London E1 4NS UK
2Natural History Museum London SW7 5BD UK 3Jodrell Laboratory Royal Botanic Gardens Kew Richmond Surrey TW93DS UK 4Max Planck Institute for Chemical Ecology Department of Molecular Ecology Beutenberg Campus Hans-Knoll-Strasse 8 07745 Jena Germany 5Institute of Biophysics Academy of Sciences of the Czech Republic CZ-61265 Brno CzechRepublic 6Institut Jean-Pierre Bourgin UMR1318 INRA-AgroParisTech INRA-Versailles 78026 Versailles cedex France
and 7Department of Plant Sciences University of Cambridge Downing Street Cambridge CB2 3EA UK For correspondence E-mail arleitchqmulacuk
dagger Present address Department of Botany and Plant Sciences University of California Riverside 900 University AvenueRiverside CA 92521 USA
Dagger Present address Natural Resources Institute University of Greenwich Chatham Maritime Kent ME4 4TB UKx Present address MRC Centre for Outbreak Analysis and Modelling Department of Infectious Disease Epidemiology
Imperial College London London W2 1NY UK
Received 6 October 2014 Returned for revision 15 January 2015 Accepted 16 March 2015
Background and Aims Speciation in angiosperms can be accompanied by changes in floral colour that mayinfluence pollinator preference and reproductive isolation This study investigates whether changes in floral colourcan accompany polyploid and homoploid hybridization important processes in angiosperm evolution Methods Spectral reflectance of corolla tissue was examined for 60 Nicotiana (Solanaceae) accessions (41 taxa)based on spectral shape (corresponding to pigmentation) as well as bee and hummingbird colour perception in orderto assess patterns of floral colour evolution Polyploid and homoploid hybrid spectra were compared with those oftheir progenitors to evaluate whether hybridization has resulted in floral colour shifts Key Results Floral colour categories in Nicotiana seem to have arisen multiple times independently during theevolution of the genus Most younger polyploids displayed an unexpected floral colour considering those of theirprogenitors in the colour perception of at least one pollinator type whereas older polyploids tended to resembleone or both of their progenitors Conclusions Floral colour evolution in Nicotiana is weakly constrained by phylogeny and colour shifts do occurin association with both polyploid and homoploid hybrid divergence Transgressive floral colour in N tabacum hasarisen by inheritance of anthocyanin pigmentation from its paternal progenitor while having a plastid phenotypelike its maternal progenitor Potentially floral colour evolution has been driven by or resulted in pollinator shiftsHowever those polyploids that are not sympatric (on a regional scale) with their progenitor lineages are typicallynot divergent in floral colour from them perhaps because of a lack of competition for pollinators
Key words Evolution floral colour hybridization Nicotiana flower pigmentation pollinator shifts polyploidySolanaceae spectral reflectance transgressive traits
INTRODUCTION
Polyploidy or whole-genome multiplication has played animportant role in the evolution of flowering plants (Soltis et al2009 2014) Allopolyploidy arising from interspecifichybridization and polyploidy can cause lsquogenomic shockrsquo(McClintock 1984) which may trigger a suite of geneticchanges including (retro)transposition differential gene ex-pression chromosome rearrangements and epigenetic changes(Leitch and Leitch 2008) These events and novel cisndashtransinteractions between progenitor genomes may generate varia-tion including transgressive phenotypes and facilitate rapid
divergence of both homoploid and allopolyploid hybrids(Wittkopp et al 2004 Chen 2007 Gaeta et al 2007 Anssouret al 2009 Tirosh et al 2009 Clare et al 2013)Speciation in angiosperms can be accompanied by or per-
haps driven by changes in floral colour that may influence pol-linator preference and reproductive isolation Many pollinatorssuch as bumblebees and hummingbirds visit a range of flowercolours (Waser et al 1996) Several species of flower-naivebumblebees have an innate colour preference for violet andblue shades although preferences in experienced foragers arelargely determined by learned associations between colours and
VC The Author 2015 Published by Oxford University Press on behalf of the Annals of Botany CompanyThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicenses
by40) which permits unrestricted reuse distribution and reproduction in any medium provided the original work is properly cited
Annals of Botany 115 1117ndash1131 2015doi101093aobmcv048 available online at wwwaoboxfordjournalsorg
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
rewards (Raine et al 2006) Hummingbirds appear to have noinnate preferences for particular colours but are likewise goodat forming associations between flower visual displays and re-wards (Goldsmith and Goldsmith 1979 Chittka and Waser1997) Hummingbirds have red receptors whereas many insectsdo not Consequently red flowers are poorly detectable by beepollinators but conspicuous for hummingbirds Thereforehummingbird-visited flowers are often red whereas those polli-nated by bees typically have a range of other colours(Rodriguez-Girones and Santamaria 2004 Shrestha et al2013) Flowers visited by nocturnal pollinators are more oftenwhite than those pollinated in full daylight probably to maxi-mize their detectability in dim light conditions (Kevan et al1996) Because of such differences in affinities of various polli-nator classes to certain flower colours differences in flowercolour can contribute to restricting gene flow between pheno-types although they will rarely result in complete reproductiveisolation for this differences in multiple traits are typically es-sential In Aquilegia (Ranunculaceae) blue- red- and whiteyellow-flowered species are primarily pollinated by bees hum-mingbirds and hawkmoths respectively (Grant 1952 Whittalland Hodges 2007) In Petunia axillaris (Solanaceae) hawk-moths prefer white flowers to pink flowers transformed to ex-press ANTHOCYANIN2 whereas bumblebees prefer pinkANTHOCYANIN2 flowers to white flowers demonstrating thatexpression of a single gene can cause differences in pollinatorvisitation (Hoballah et al 2007) Similarly manipulation of asingle locus controlling carotenoid production inMimulus flow-ers (Phrymaceae) results in a pollinator shift reaffirming theimportance of floral colour in determining pollinator behaviour(Bradshaw and Schemske 2003)To analyse floral colour in the context of pollination it is
necessary to consider both colour theory and pollinator visualsystems There are several important differences between thecolour vision systems of humans and those of various pollinatortypes Humans and many insects are trichromatic havingthree discrete photoreceptor types however humans possessred- (with peak sensitivity (kmax) near 560 nm) green-(kmaxfrac14 535 nm) and blue-sensitive (kmaxfrac14 420 nm) photore-ceptors (Bowmaker and Dartnall 1980) whereas many insectshave ultraviolet- (UV- kmax 350 nm) blue- (kmax 440 nm)and green-sensitive (kmax 530 nm) receptors (Peitsch et al1992 Briscoe and Chittka 2001 Kelber et al 2003) Manybirds (Bowmaker 1998) and some butterflies (Kelber 2001)have tetrachromatic colour vision In birds photoreceptors aresensitive to red green blue and either violet or UV wave-lengths (Hart and Hunt 2007) Hummingbirds have four singlecone types with peak sensitivities in the UV (kmaxfrac14 370 nm)blue (kmaxfrac14 440 nm) bluendashgreen (kmaxfrac14 508 nm) and yellow(kmaxfrac14 560 nm) the sensitivity of the last extends significantlyinto the red (Herrera et al 2008) Hummingbirds can learn todistinguish near-UV light (370 nm) from darkness confirmingthat they use their UV receptors for colour vision at a behaviou-ral level (Goldsmith 1980) We will take into account these dif-ferences in pollinator perception as we examine floral colour inthe genus Nicotiana (Solanaceae)We investigate the evolution of floral colour across
Nicotiana (Solanaceae) in the context of polyploidy and hybrid-ization Nicotiana is an excellent group in which to study the ef-fects of hybridization as nearly half of the 76 species are
allotetraploids of different ages (Chase et al 2003Clarkson et al 2004 2005 Leitch et al 2008 Kelly et al2013) and several putative homoploid (diploid) hybrids havealso been detected (Clarkson et al 2010 Kelly et al 2010)which add to the reticulate nature of the genus These phyloge-netic studies have also been used to predict the closest livingdescendent species of the parents that formed the homoploidhybrid and allopolyploid species hereafter called progenitorspecies as shown in Fig 1 Some synthetic polyploidsmade from these progenitor species are also availableproviding insight into the immediate effects of polyploidy andhybridization We compare floral colours of Nicotianapolyploid and homoploid hybrids with those of their diploidprogenitorsBecause various animal groups have different sensitivities
to colour it is necessary to model colour perception ofspecific pollinator classes to understand the significance offloral colour signals Here we consider floral colours from abee perspective (Chittka 1992) which can also be used to rep-resent other trichromatic insects such as hawkmoths due tosimilarities in photoreceptor sensitivities (Kelber et al 2003)and a hummingbird perspective (Herrera et al 2008 Restrepo2013) Hummingbirds and hawkmoths are known tovisit Nicotiana species (Aigner and Scott 2002 Kaczorowskiet al 2005 Kessler and Baldwin 2006 Nattero and Cocucci2007)Pigments typically determine floral colour however few
studies have examined the specific pigments present inNicotiana petals Aharoni et al (2001) confirm the presence ofanthocyanin pigmentation in N tabacum which seems to bepredominantly cyanidin derivatives Spectral colour shifts canoccur in anthocyanins due to hydroxylation and methylationwhich result in different types of anthocyanins (Castaneda-Ovando et al 2009 Andersen and Jordheim 2010) and differ-ences in pH as well as copigmentation with other pigmentsincluding carotenoids and colourless flavonoids or metal ionscan also cause spectral shifts in the same anthocyanincompound (Grotewold 2006 Andersen and Jordheim 2010)The yellow flower colour of Nicotiana glauca is due to caroten-oid pigmentation (Zhu et al 2007) Crossing experimentsbetween diploid Nicotiana species suggest that the presence ofchlorophyll in corolla tissue is dominant (Brieger 1935)and similar results corroborate this in the carnation Dianthuscaryophyllus (Caryophyllaceae Ohmiya et al 2014)In this paper we seek to determine what types of spectral
reflectance are found within Nicotiana and how they appear tobee and hummingbird pollinators We focus on the conse-quences of polyploidy and interspecific hybridization on floralcolour evolution Specifically we aimed to test the hypothesesthat (1) polyploid and homoploid hybrids will have floralcolours that will resemble at least one of their progenitors indiscrete spectral bee and hummingbird floral colour categoriesobtained from cluster analyses (2) polyploid and homoploidhybrids will be positive for chlorophyll pigmentation in corollatissue if at least one progenitor has chlorophyll present in itspetals (due to evidence of the dominance of chlorophyll pig-mentation) (3) increased cell size potentially associated withpolyploidy affects the concentration of pigments and in turncolour intensity and (4) floral colour evolution is constrainedby phylogeny
1118 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
N tabacum 095-55
N tabacum 51789
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N sylvestris N tomentosiformis
N tabacum lt0middot2 myo
times
Natural Synthetic
A
timesN sylvestris N otophora TH32
TH32 synthetic polyploid
Synthetic
B
timesN rustica
var asiaticaN rustica
var pavoniisynthetic
UxPsynthetic PUE1 F1
synthetic N rusticaPUE1-R10
S0
synthetic N rustica
PUE1-R1 S1
N paniculata N undulata
N rustica lt0middot2 myo
Natural
Synthetic
C
timesN undulata N wigandioides N arentsii
N arentsii lt0middot2 myo
Natural
D
Natural Synthetic
timesN clevelandii N quadrivalvis
904750042N quadrivalvis
TW18N obtusifolia
lsquoBaldwinrsquo
N attenuata
lsquoBaldwinrsquo
N times obtusiata line 1
N times obtusiata line 2
N times obtusiata line 5
Section Polydicliae ~1 myoE
timesN repanda N nesophila N stocktonii N nudicaulisN sylvestris N obtusifolia
TW143
Section Repandae~4middot5 myoF
Natural
NoctifloraendashPetunioides homoploid hybrids
N noctiflora N petunioides N acuminata N attenuata N miersii N pauciflora N linearis TW77
N glauca 51725
N glauca 51751
times
Section PetunioidesSection Noctiflorae
H
N glutinosa
N otophora N setchellii N tomentosiformis N undulata N wigandioides N glutinosa
times
Section Tomentosae Section Undulatae
I
Section Suaveolentes ~10 myo
N noctiflora N petunioides N acuminata N attenuata N miersii N pauciflora
times
G
Section PetunioidesSection Noctiflorae
times
N sylvestris
N benthamiana
N forsteri
N gossei
N occidentalis
var hesperis
N megalosiphonN suaveolens
FIG 1 Floral colour as perceived by humans of polyploid and homoploid hybrid Nicotiana and their diploid progenitors Polyploid ages were estimated using a mo-lecular clock calibrated with the geological age of volcanic islands with endemic Nicotiana species (Clarkson et al 2005) Absolute dates (millions of years myo)estimated by the clock should be treated with caution however relative ages of different polyploid sections should reflect the true sequence of polyploidizationevents (A) Natural and synthetic polyploids of N tabacum (B) Synthetic polyploid TH32 (C) Natural and synthetic N rustica polyploids Synthetic hybrids includea homoploid from a reciprocal cross and a polyploid series (F1 homoploid and S0 and S1 polyploids) of the same parentage as natural N rustica (D) Nicotiana are-ntsii (E) Natural polyploids of section Polydicliae Synthetic N obtusiata polyploid lines were made from a cross between the N obtusifolia and N attenuata ac-cessions studied here (F) Section Repandae (G) Section Suaveolentes contains 26 polyploid species (six included in this study) Biogeographical analyses suggestthat section Suaveolentes originated15 million years ago (mya) before the aridification of Australia (Ladiges et al 2011) and this seems to be relatively congru-ent with the molecular clock results which place its origin at 10mya (H) Homoploid hybrids N glauca and N linearis (I) Homoploid hybrid N glutinosa
Photographs are scaled to the same size
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1119
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
MATERIALS AND METHODS
Petal cell area measurements
To assess whether an increase in ploidy results in larger petalcells cell area was measured from a subset of polyploids andtheir progenitors For Nicotiana sylvestris A04750326Nicotiana rustica var asiatica Nicotiana rustica var pavoniiNicotiana paniculata Nicotiana undulata and Nicotiana nudi-caulis mature flowers were taken from plants and the adaxialpetal surface was imprinted in Elite HD vinylpolysiloxane im-pression material (dental wax supplied by ZhermackHarrogate UK) The wax was left to set and then used as amould for making epoxy petal casts Devcon high-strength ep-oxy was mixed according to the manufacturerrsquos instructionspoured into the mould and allowed to set for 12 h The epoxyrelief was removed and coated with gold using a QuorumK756X sputter coater The samples were then imaged using aFEI Philips XL30 FEGSEM scanning electron microscope ForNicotiana obtusifolia var obtusifolia TW143 Nicotianarepanda and Nicotiana stocktonii only fixed material wasavailable whole mature flowers were fixed in formalinndashaceticacidndashalcohol (FAA) (60 ethanol 6 formaldehyde 5 acetic acid) for 72h before being transferred to a 70 ethanol(EtOH) wash for 24 h The samples were then dehydratedthrough an ethanol series of 2 h each in 70 80 and 90 andtwo washes in 100 EtOH The samples were dissected andthen dried in an Autosamdri 815B critical point dryer Thesesamples were sputter-coated and imaged as described aboveFor all samples images were taken mid-petal from an angleperpendicular to the surface to minimize parallax error Cellsize measurements were carried out in ImageJ (httpimagejnihgovij) The circumference of the cell base was drawn free-hand and area was calculated for 100ndash150 cells until the cu-mulative mean stabilized One-way ANOVA and Tukeyrsquoshonest significance tests were performed in RStudio version098490 (httpwwwrstudioorg) to compare cell area of poly-ploids with those of their progenitors repeating the tests foreach polyploid section
Spectral reflectance measurements
Spectral reflectance measurements were recorded for 60Nicotiana accessions (41 taxa Supplementary Data Table S1)three flowers from different plants where possible were usedfor each accession Reflectance spectra from three Nicotianaotophora accessions were pooled because the spectra weresimilarSpectral reflectance of flowers at anthesis was measured
from 300 to 700 nm using an Avantes AvaSpec-2048 spectro-photometer with an Avantes AvaLight-DHS light source andcalibrated with a barium sulphate white standard from lab-sphereVR Nicotiana mutabilis was also measured later as flow-ers change from white to pink when mature pink flowers areless likely to have a nectar reward but add to the attraction ofthe overall floral display and therefore are still relevant to pol-linators (R Kaczorowski University of Haifa Israel perscomm) Reflectance spectra express the proportion of lightreflected by the flower at any given wavelength Spectra werevisualized and exported in increments of 1 nm using the
program AvaSoft version 703 Full (Avantes BV Eerbeek TheNetherlands) and imported into ExcelSpectra for each accession or colour morph were averaged
and then smoothed three times using a rolling average over9 nm Spectra for all accessions were submitted to the FloralReflectance Database (FReD wwwreflectancecouk Arnoldet al 2010)Some spectra had a spike at 656 nm which corresponded
to a narrow peak in the light source spectrum suggesting thatthe spectra were saturated at 656 nm however smoothingserved to neutralize this spike Several spectra (Nicotiana are-ntsii N mutabilis Nicotiana suaveolens and Nicotiana wigan-dioides) included an anomalous reflectance minimum from 475to 500 nm which could not be explained by the light sourcespectrum Remeasured spectra of N arentsii N suaveolens andN wigandioides lacked this minimum but further material ofN mutabilis was unavailable so these spectra were includeddespite the anomalies
Calculation of colour loci in the bee colour hexagon
A reflectance spectrum can be represented as a single pointin the bee colour hexagon space (a graphical representation of abeersquos colour visual experience) based on the relative excitationof UV- blue- and green-sensitive photoreceptor types (Chittka1992) Vertices of this hexagon represent theoretical stateswhere one or two photoreceptor types are at maximal excitationwhereas at least one receptor type is at zero excitation (eg thetop vertex of the hexagon corresponds to maximal blue receptorexcitation and zero signal from UV and green receptorswhereas the top right vertex corresponds to maximal signal inboth blue and green receptors but no signal in the UV receptorand so forth see Supplementary Data Fig S1) The centre ororigin of the hexagon is achromatic Hue corresponds to angu-lar position around the origin whereas spectral purity or satura-tion increases with distance from the originBee colour hexagon coordinates were calculated for all
Nicotiana spectra Illumination was assumed to be sunlight(D65 Wyszecki and Stiles 1982) the background was repre-sented by an average leaf spectrum (Gumbert et al 1999)Honeybee photoreceptor spectral sensitivity functions wereused to determine bee colour hexagon coordinates these aresimilar to bumblebee and hawkmoth photoreceptor sensitivityfunctions so the bee colour hexagon can be used to approxi-mate the colour vision of these insects as well (Menzel et al1986 Peitsch et al 1992 Briscoe and Chittka 2001 Kelberet al 2003 and references therein Skorupski et al 2007) Theequations used to determine colour hexagon coordinates are asfollows where EG EB and EUV represent the excitation of thegreen blue and UV bee photoreceptors respectively elicitedby a spectrum (Chittka 1992)
x frac14ffiffiffi
3p
=2ethEG EUVTHORNy frac14 EB 05ethEUV thorn EGTHORN
Because the colour loci of Nicotiana flowers were mostlyclose to the centre of the colour space all colour hexagon dis-plays presented are scaled so that only the central 40 is
1120 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
shown the outline is therefore drawn as a dashed line This re-sults in a clearer spread of the colour loci to facilitate visual in-spection For a diagram explaining the colour hexagon seeSupplementary Data Fig S1
Calculation of colour loci in hummingbird colour space
For tetrachromatic hummingbirds we chose to model flowercolours in a 3-D colour opponent space because n ndash 1 colouropponent dimensions are necessary to code the informationfrom n colour receptors (Chittka 1996) The hummingbirdcolour space can be displayed as a rhombic dodecahedron with14 vertices (Restrepo 2013) Like the bee colour hexagonvertices of the space represent states where one two or threephotoreceptor types are at maximal excitation and at least onereceptor type is at zero excitation
Hummingbird colour space coordinates were calculated forall Nicotiana spectra Illumination was again assumed to besunlight (D65 Wyszecki and Stiles 1982) and the backgroundan average leaf spectrum (Gumbert et al 1999) as was usedfor the bee colour hexagon Photoreceptor spectral sensitivityfunctions from the green-backed firecrown hummingbird(Sephanoides sephanoides Herrera et al 2008) were used todetermine hummingbird colour space coordinates usingthe following equations (Restrepo 2013) where ER EGEB and EUV represent excitation of red green blue andUV hummingbird photoreceptors respectively elicited by aspectrum
x frac14ffiffiffiffiffiffiffiffi
3=4p
EB ffiffiffiffiffiffiffiffiffiffi
1=12p
ethEUV thorn EG thorn ERTHORN
y frac14ffiffiffiffiffiffiffiffi
2=3p
EG ffiffiffiffiffiffiffiffi
1=6p
ethEUV thorn ERTHORN
z frac14ffiffiffiffiffiffiffiffi
1=2p
ethEUV ERTHORN
RStudio was used to make 3D plots of the hummingbird col-our space and ImageJ version 148 (httpimagejnihgovij)was used to create an animation of the Nicotiana flower loci inthe hummingbird colour space Again Nicotiana flower colourloci are close to the origin in the hummingbird colour space sothe graphs presented display only the central portion (either 25or 50 ) of the colour space for clarity To further facilitateinterpretation of these graphs vertices representing individualexcitation of the red green blue and UV photoreceptor typesas well as their excitation vectors from the origin are shownin red green blue and black respectively Other vertices(representing excitation of two or three photoreceptor types) areshown in grey
Cluster analyses
Cluster analyses were used to group spectra based on spectralshape (corresponding to pigmentation) and their position inboth bee and hummingbird colour spaces For spectral colourcategories spectra were normalized to the same integral underthe curve in order to compare combinations of pigments notthe concentration of pigments A distance matrix was calculatedfrom the normalized spectral data in R version 302 (http
wwwR-projectorg) using the dist() function For the bee andhummingbird colour categories the input data were the (x y) or(x y z) coordinates of the spectra in the bee and hummingbirdcolour spaces respectivelyData were first imported into R The function hclust() was
used to perform agglomerative hierarchical clustering based onthe average pairwise distances between groups With this algo-rithm the observed points which are initially all deemed to bedistinct are iteratively assigned to groups until eventually allpoints belong to the same group At each step the average dis-tance between all groups (ie the mean distance from all pointsin group A to all points in group B if either one of these is asingle point then no averaging is needed) is calculated and thetwo groups with the smallest average distance are merged Theorder in which groups are merged can be used to construct adendrogram showing the spatial relationship between all datapoints We can also look at the distribution of merge distancesat each step in the algorithm and use this distribution to esti-mate how many groups are present in the data Points at whichthere is a steep increase in the average between-group distance(lsquoelbowrsquo points) highlight the spatial scale(s) at which there isclustering present in the data By using one of these elbowpoints as a cutoff in the algorithm we can arrive at a distancegrouping that captures the spatial clustering It should be notedthat the determination of where to draw the threshold in a clus-ter analysis is arbitrary but the use of one of these elbow pointsdoes yield meaningful clusters The determination of the spe-cific point from the elbow region to be used to define clusterswas further informed by visual inspection of reflectance spec-tra as well as distributions of colour loci in the perceptual col-our spaces It should be noted that the dendrograms relate tosimilarities in spectral reflectance as well as colour relation-ships perceived by bees and hummingbirds they do not showphylogenetic relationships
Ancestral state reconstruction
To examine the evolution of colour within a phylogeneticcontext ancestral state reconstructions were performed on treesinferred from plastid sequence data Only species for which flo-ral character data are available were included in these analysesBecause polyploid and homoploid hybrids originate via reticu-late evolutionary processes and therefore lack a history oftree-like evolution ancestral characters were reconstructed us-ing only non-hybrid diploid species The states observed inhybrid species were then compared with the ancestral state re-constructions Since sections Repandae and Suaveolentes havediversified to form several species following polyploidizationcharacters were reconstructed for these sections separately toexamine colour evolution subsequent to their origin Fornon-hybrid diploid species individual gene trees yieldsome conflicting topologies nevertheless key nodes for thepurposes of interpreting character evolution in hybrids arerecovered in multiple gene trees and are supported bysupernetwork analyses (Kelly et al 2010) Therefore plastiddata from previously published studies are suitable for theseanalysesPreviously published sequences (Clarkson et al 2004) from
four plastid regions (matK ndhF trnL-F and trnS-G) were
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1121
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
aligned separately using PRANKthornF (Loytynoja and Goldman2008) and then concatenated to create a combined plastid data-set before further optimization by eye in Mesquite version 274(Maddison and Maddison 2008) For Nicotiana attenuata weused GenBank accessions AB040009 and AY098697 for thematK and trnL-F regions respectively (due to likely misidenti-fication of N attenuata material used in Clarkson et al 2004see Clarkson et al 2010) the other two regions were scored asmissing data Phylogenetic reconstruction by Bayesian infer-ence was performed as described in Kelly et al (2013) with theexception that BayesTrees v13 (wwwevolutionreadingacukBayesTreeshtml) was used to construct 95 majority ruleconsensus trees For sections Repandae and Suaveolentes se-quences representing their putative maternal progenitors wereincluded during Bayesian inference to allow rooting of trees butwere pruned from the trees prior to ancestral statereconstructionAncestral states for spectral reflectance colour categories and
presenceabsence of chlorophyll in petals (data inSupplementary Data Table S2) were reconstructed using theparsimony reconstruction method in Mesquite version 274 un-der the unordered states assumption To account for topologicaluncertainty character states were reconstructed over all 36 000post-burn-in trees using the Trace Character Over Trees optionand summarized on the 95 majority rule consensus tree fromthe Bayesian analysis Ancestral states were not calculated forbee or hummingbird colour categories because these are per-ceptual systems and the same colour category can result fromdifferent combinations of pigments (eg both human pink andhuman white flowers which are positive and negative respec-tively for anthocyanin pigmentation are both classified as beebluendashgreen) thus a single colour category does not necessarilyhave a shared evolutionary history
Estimating expected polyploid and homoploid hybrid floralcolour
Polyploid and homoploid hybrid floral colours for each ac-cession were compared with those of their diploid progenitorsfor the spectral bee and hummingbird colour categories definedby cluster analyses Floral colour was classified as lsquoexpectedrsquo ifit fell in the colour category of at least one progenitor or lsquounex-pectedrsquo if it was different from both progenitors Polyploid andhomoploid hybrids were also compared with their diploid pro-genitors for the presence or absence of chlorophyll in corollatissue Chlorophyll absorbs at 675 nm in vivo (Haardt andMaske 1987) therefore the presence of chlorophyll can be in-ferred from reflectance spectra if there is a reflectance mini-mum at 675 nm The presence of chlorophyll in petal tissueappears to be dominant (Brieger 1935 Ohmiya et al 2014)Thus hybrids were classified as expected if they showed chlo-rophyll pigmentation and as unexpected if they did not becauseat least one progenitor possessed chlorophyll in all diploid pro-genitor combinations For natural homoploid hybrids and poly-ploid section Suaveolentes where progenitors can only bedefined to Nicotiana section level comparisons were madewith reconstructed ancestral characters thus only spectral col-our categories and the presenceabsence of chlorophyll wereexamined
Phylogenetic signal in floral traits
In order to statistically test for phylogenetic signal in the phe-notypic trait data (spectral reflectance bee and hummingbirdcolour perception) we used Mantel tests to examine the corre-lation between phylogenetic distance and each of the respectivecontinuous multidimensional traits (eg Cubo et al 2005Muchhala et al 2014) Analyses were restricted to diploid spe-cies excluding homoploid and polyploid hybrids Trees wereedited in Newick format to include additional tips with zerobranch lengths for taxa that are multiple in the trait datasetseither due to colour polymorphism (N otophora) or multipleaccessions (N sylvestris and N obtusifolia var obtusifolia)Statistical analyses were performed in R version 310
Phenotypic distance matrices were first calculated for the threetrait datasets using Euclidean distance and phylogenetic dis-tance matrices were calculated (1) as genetic distance from theplastid alignment and (2) for each of 36 000 post-burn-inBayesian trees using copheneticphylo() part of the APE pack-age version 31-2 (Paradis et al 2004) The second Bayesianset of tests was performed in order to account for evolutionaryprocesses such as saturation and to estimate how phylogeneticuncertainty affects the correlation Mantel tests were performedusing Pearsonrsquos product-moment correlation coefficient with10 000 permutations of each distance matrix to test for signifi-cance the mean P value and its standard deviation were calcu-lated for each set of 36 000 Mantel tests from the Bayesiantrees along with the percentage of trees that gave significantcorrelations The function mantel() from the vegan packagewas used (Oksanen et al 2013)
RESULTS
Petal cell area
Petal cell area was measured to determine whether an increasein ploidy results in larger floral cells Polyploid petal cell areawas significantly larger than in both progenitors in N rustica(ANOVA Ffrac14 371 dffrac14 3 Plt 2 10ndash16) accessions but wasintermediate between progenitors in section Repandae poly-ploids (ANOVA Ffrac14 2492 dffrac14 4 Plt 2 10ndash16 Fig 2)Accessions that were significantly different in cell area (withinpolyploid sections and their progenitors) are represented bydifferent letters above the bars in Fig 2 results from Tukeyrsquoshonest significance tests can be found in Supplementary DataTable S3
Cluster analyses
Nicotiana reflectance spectra were grouped into categoriesbased on spectral shape and position in the bee and humming-bird colour spaces using cluster analyses Bees and humming-birds have different photoreceptor sensitivities and we expectour cluster analyses to reflect these differences in sensoryequipment The analysis based on spectral shape yielded eightcolour categories which roughly corresponded to flowers per-ceived by human observers as magenta red pink UVndashwhitewhite yellow green and dark green (Fig 3) Nicotiana spectraare displayed by spectral colour category in Fig 4A BSupplementary Data Fig S2 The bee colour hexagon clustering
1122 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
resulted in 11 colour categories which fell into the followingareas of bee colour space saturated green UVndashblue high UVUVndashgreen green light green bluendashgreen dark green saturatedUVndashblue saturated UVndashgreen and blue (the last four categorieswere each represented by only a single accessionSupplementary Data Fig S3A) These groups are shown in thebee colour hexagon (Fig 4C) The hummingbird colour spacecluster analysis also produced 11 colour categories saturatedgreen green UVndashwhite UVndashgreen pink white UVndashpinkdark green light pink red and saturated UVndashpink (again thelast four categories include only a single accessionSupplementary Data Fig S3B) These groups are shown in thehummingbird colour space (Fig 4D) and the same graph isprovided as an animation to better display the 3D nature of thecolour space (Supplementary Data Video)
Evolution of spectral reflectance in polyploids and homoploidhybrids
To assess the evolution of polyploid floral colour polyploidand homoploid hybrid accessions were compared with those oftheir progenitors in spectral bee and hummingbird colour cate-gories as well as in the presenceabsence of chlorophyll Thediploid progenitors and approximate age of polyploids andhomoploid hybrids are found in Fig 1 and Supplementary DataTable S4 and the observed and expected floral colours of
polyploid and homoploid hybrids are found in Table 1 Mostpolyploid and homoploid hybrids were similar to at least oneprogenitor in spectral bee and hummingbird colour categoriesbut some fell into unexpected colour categories (Table 1Fig 5 Supplementary Data Fig S4 Supplementary Data FigS5) Over half of the polyploids unexpectedly lacked chloro-phyll (Table 1)
Evolution of colour characters in a phylogenetic context
Reconstructed character states are shown for spectral reflec-tance colour categories (Fig 6) and the presenceabsence ofchlorophyll in petals (Supplementary Data Fig S6) Bee andhummingbird colour categories are also shown for extant spe-cies on the plastid tree (Fig 6) Although the deepest nodeswere largely equivocal evolution of spectral reflectance colourin Nicotiana seemed to be dynamic (Fig 6) Green flowerslikely have three independent origins (1) in sectionsPaniculatae and Undulatae (2) in N langsdorffii and (3) inthe homoploid hybrid N glauca UVndashwhite flowers also seemto have arisen three times independently (1) in sectionTrigonophyllae (2) in N pauciflora and (3) in the homoploidhybrid N linearis Most polyploid and homoploid hybrid spe-cies exhibit a floral colour present in at least one of their pro-genitors However N tabacum 095-55 is red and N glauca isyellow and green unlike their progenitors UVndashwhite flowersseem to have evolved de novo in N linearis UVndashwhite flowersare also found in one of its progenitor sections (in N pauci-flora) but ancestral reconstructions indicate that the floral col-our was most likely white at the ancestral nodes within thesection (Fig 6) This suggests that the evolution of UVndashwhiteflowers in N pauciflora has occurred subsequent to the forma-tion of N linearis and that the two events are likely indepen-dent It is unclear whether UVndashwhite flowers also evolved denovo in N nudicaulis because the ancestral node of sectionRepandae is equivocal The presence of chlorophyll as inferredby light absorption at 675 nm (Haardt and Maske 1987) inNicotiana flowers is ancestral and has been lost three times inN sylvestris N noctiflora and the most recent common ances-tor of N acuminata and N pauciflora (Supplementary DataFig S6)Results from Mantel tests of phylogenetic signal for
Nicotiana floral traits for both genetic distance and the 36 000post-burn-in Bayesian trees are shown in Table 2 All floraltraits were significantly correlated with phylogenetic relation-ships for the Bayesian trees at a significance level of Plt 005Only spectral reflectance was significant for the genetic dis-tance tests whereas bee and hummingbird colour perceptionwere just above the Plt 005 threshold For the Bayesian trees901 662 and 932 of trees were significantly correlatedwith the spectral reflectance bee and hummingbird colour per-ception datasets respectively These results suggest that thesefloral traits are weakly constrained by phylogeny
DISCUSSION
Nicotiana is remarkable in its range of spectral reflectanceflower colours (white UVndashwhite pink magenta red yellowgreen and dark green Fig 3) and in the variety of pollinators
0
200
400
600
800
1000
1200C
ell
are
a (
microm
2)
N rustica
N paniculata
N undulata
N sylvestris
N obtusifolia
N nudicaulis
N repanda
N stocktonii
N rustica var asiatica
N rustica var pavonii
Repandae
a
a
b
b
cc
c c
d
Maternal progenitor
Paternal progenitor
Polyploids
Polyploids
Polyploids
FIG 2 Petal cell area from polyploids and their progenitors Within each poly-ploid group bars with different letters represent significantly different mean cell
areas
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1123
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
that visit the flowers (moth bird bee bat Knapp 2010) Theperception of these spectral colours also changes with visualsystem (bee or hummingbird) Here we describe a complex dy-namic in the evolution of floral colour in Nicotiana Spectral re-flectance and bee and hummingbird colour perception arecorrelated with phylogeny but multiple independent origins ofvarious combinations of pigmentation suggest that the evolutionof floral colour is not entirely phylogenetically constrained
Petal cell size evolution in polyploids
Cell size is expected to increase following polyploidizationdue to the increase in genome size (Beaulieu et al 2008) Thesignificantly larger petal cells of N rustica (lt02 million yearsold myo) and the intermediate cell size of section Repandae(45myo) suggest that polyploids may revert to a diploid-like cell size over time similar to the genome downsizing ob-served in polyploids (Leitch and Bennett 2004) Howeverpetal cell size differences within section Repandae do not seemto be linked to genome size N nudicaulis and N repanda sharesimilar cell sizes but have substantially different genome sizes(Leitch et al 2008) The concentration of pigment in floralcells is controlled by both the amount of pigment present andthe cell size Nicotiana rustica has significantly larger petalcells than its progenitors (close to the sum Fig 2) and displaysan intermediate brightness (the area under the reflectance curvea proxy for pigment concentration) between its progenitors
(Supplementary Data Fig S4D) This is expected if the poly-ploid inherits the sum of both cell size and amount of pigmentpresent from its progenitors
Polyploid divergence in floral colour
Many younger polyploids (lt02myo) display unexpectedfloral colours considering those of their diploid progenitorsNone of the natural and synthetic N tabacum accessions pos-sess chlorophyll which is unexpected given its presence in atleast one progenitor species Nicotiana tabacum 095-55 alsohas unexpected spectral bee and hummingbird colour giventhe colour categories of the progenitor species Similarly syn-thetic N tabacum QM has unexpected colours in bee and hum-mingbird perception Because this accession is synthetic theparents are known and thus its unexpected phenotype can beclassified as transgressive or outside the range of its progeni-tors due solely to polyploidy and hybridization Most N rusticaaccessions have unexpected bee colour (four of these aresynthetic and are therefore also transgressive) and N arentsiihas unexpected hummingbird colour (Table 1 Fig 6) Despitethe divergence of floral spectra associated with polyploidybehavioural studies are needed to determine whether the beeand hummingbird colour categories delineated here actuallyelicit different responses in pollinatorsMost older polyploids (1ndash10myo) are similar in floral col-
our category to at least one of their progenitors N clevelandii
synth
etic
N ru
stic
a P
UE
1-R
1 S
1
N ru
stic
a v
ar a
sia
tica
synth
etic
N ru
stic
a P
UE
1-R
10 S
0
synth
etic
PU
E1 F
1synth
etic
UxP
N b
enavid
esii
N ru
stic
a v
ar p
avonii
N p
anic
ula
taN
langsdorffii
N g
lauca 5
1751 g
reen
N u
ndula
taN
raim
ondii
N k
nig
htia
na
N g
lauca 5
1751 y
ello
w
N g
lauca 5
1725
N p
lum
bagin
ifolia
N g
ossei
N q
uadriv
alv
is T
W18
N o
ccid
enta
lis s
ubsp h
esperis
N
megalo
sip
hon
N s
uaveole
ns
N n
esophila
N s
tockto
nii
N q
uadriv
alv
is 9
04750042
N a
ttenuata
N re
panda
N s
ylv
estris
A04750326
N
benth
am
iana
N m
uta
bilis
CP
G3 w
hite
N
muta
bilis
CP
G12456 w
hite
N
wig
andio
ides
N a
cum
inata
N fo
rste
riN
cle
vela
ndii
N n
octiflo
raN
x o
btu
sia
ta lin
e 2
N
x o
btu
sia
ta lin
e 1
N
mie
rsii
N p
etu
nio
ides
N x
obtu
sia
ta lin
e 5
N
sylv
estris
6898
N
oto
phora
white
N
are
nts
iiT
H3
2N
tabacum
lsquoChulu
manirsquo
N o
btu
sifo
lia v
ar p
alm
eri
N o
btu
sifo
lia v
ar o
btu
sifo
lia lsquoB
ald
win
rsquoN
linearis
TW
77
N n
udic
aulis
N o
btu
sifo
lia v
ar o
btu
sifo
lia T
W143
N lin
earis
964750099
N
pauciflo
raN
tom
ento
sifo
rmis
N o
tophora
pin
k
N g
lutin
osa
synth
etic
N ta
bacum
TH
37
N
tabacum
51789
synth
etic
N ta
bacum
QM
N
tabacum
095-5
5
N s
etc
hellii
N m
uta
bilis
CP
G3 p
ink
N m
uta
bilis
CP
G12456 p
ink
10
8
6
4
2
0
Dis
tan
ce
Spectral categories
MagentaRedPinkUV-white
WhiteYellowDark greenGreen
Dark greenSaturated greenGreenUV-whiteUV-green
PinkLight pinkWhiteUV-pinkRedSaturated UV-pink
SpecBeeHum
SpectralDark greenSaturated greenSaturated UV-blueUV-blueSaturated UV-greenHigh UV
UV-greenGreenLight greenBlueBlue-green
Bee Hummingbird
FIG 3 Dendrograms based on distance cluster analyses for spectral reflectance Coloured circles on the dendrogram represent distinct colour categories as determinedby the chosen threshold (dashed line) Dendrograms were similarly constructed for bee and hummingbird colour (see Supplementary Data Fig S3) The lines of col-oured circles at the tips of the dendrograms signify the category each taxon is assigned to in spectral bee and hummingbird colour as labelled (Spec Bee and Hum)for comparison between spectral categories and those of different visual systems Diploids polyploids and homoploids are denoted by black blue and orange text
respectively
1124 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
0
0middot2
0middot4
0middot6
0middot8
1middot0
Reflecta
nce
Wavelength (nm)
0
0middot2
0middot4
0middot6
0middot8
1middot0
300 350 400 450 500 550 600 650 700 300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
N benavidesii
N glauca 51751 g
N langsdorffii
N paniculata
N raimondii
N rustica var asiatica
N rustica var pavonii
synthetic UxP
synthetic F1
syn N rustica S0
syn N rustica S1
N undulata
N glutinosa
N otophora p
N tabacum 51789
syn N tabacum QM
syn N tabacum TH37
N tomentosiformis
A B
G 50
R 50
B 50
UV 50
Hummingbird colour space
White
0middot40middot4
ndash0middot40middot4
0middot4
0middot4
y
z
UV-white
Pink
Synthetic N tabacum QM
UV-pink
N tabacum 095-55
N setchellii
Green
N langsdorffii
Saturated green
UV-green
C DBee colour hexagon
Blue-green
Light green
Green
Dark green
Sat green
UV-green
Sat UV-green
B
B-G
G
UV-G
UV
UV-B
1 acum2 aren3 atten4 bena5 benth6 clev7 forst8 glau259 glau51y10 glau51g11 glut12 goss13 knight14 langs15 lin964716 linTW7717 mega18 mier19 mut1w20 mut1p21 mut3w22mut3p23 neso24 noct25 nudi26 xobtus127 xobtus228 xobtus529 obtusB30 obtusTW31 obtuspalm
32 occhesp33 otoph w34 otoph p35 pani36 pauc37 petun38 plumba39 quad904740 quadTW1841 raim42 repa43 rustasi44 rustpav45 syn UxP46 syn F147 syn rust S048 syn rust S149 setch50 stock51 suav52 sylv689853 sylvA04754 tab0955555 tab5178956 tabchulu57 syntabQM58 syntabTH3759 tomtform60 undu61 wigan62TH32
High UV
UV-blue
Sat UV-blue
Blue
FIG 4 (A B) Nicotiana reflectance spectra from 300 to 700nm which roughly correspond to colours perceived by human observers as pink (A) and green (B) SeeSupplementary Data Fig S2 for other spectral colour categories Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploidhybrid taxa Abbreviations p pink syn synthetic g green (C) Colour hexagon displaying the distribution of Nicotiana colour loci in bee colour space The hexa-gon has been scaled so that vertices represent 40 excitation of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-GUVndashgreen Bee colour categories are delineated by coloured ovals sat saturated Nicotiana species abbreviations are as follows acum acuminata aren arentsiiatten attenuata benavid benavidesii benth benthamiana clev clevelandii forst forsteri glau25 glauca 51725 glau51y glauca 51751 yellow glau51g glauca51751 green glut glutinosa goss gossei knight knightiana langs langsdorffii lin9647 linearis 964750099 linTW77 linearis TW77 mega megalosiphon miermiersii mut1w mutabilis CPG12456 white mut1p mutabilis CPG12456 pink mut3w mutabilis CPG3 white mut3p mutabilis CPG3 pink neso nesophila noctnoctiflora nudi nudicaulis obtus1 obtusiata line 1 obtus2 obtusiata line 2 obtus5 obtusiata line 5 obtusB obtusifolia var obtusifolia lsquoBaldwinrsquoobtusTW obtusifolia var obtusifolia TW143 obtuspalm obtusifolia var palmeri occhesp occidentalis subsp hesperis otoph w otophora white otoph p otophorapink pani paniculata pauc pauciflora petun petunioides plumba plumbaginifolia quad9047 quadrivalvis 904750042 quadTW18 quadrivalvis TW18 raimraimondii repa repanda rustasi rustica var asiatica rustpav rustica var pavonii syn UP synthetic UP syn F1 synthetic PUE1 F1 synrust S0 synthetic rus-tica PUE1-R10 S0 synrust S1 synthetic rustica PUE1-R1 S1 setch setchellii stock stocktonii suav suaveolens sylv6898 sylvestris 6898 sylvA047 sylvestrisA04750326 tab09555 tabacum 095-55 tab51789 tabacum 51789 tabchulu tabacum lsquoChulumanirsquo syntabQM synthetic tabacum QM syntabTH37 synthetictabacum TH37 tomtform tomentosiformis undu undulata wigan wigandioides TH32 TH32 synthetic N sylvestrisN otophora polyploid (D) Distribution ofNicotiana spectral loci in hummingbird colour space Vertices of the hummingbird colour space represent 50 excitation of the photoreceptors single photoreceptortype vertices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows repre-sent the vectors of these photoreceptors from the origin of the hummingbird colour space Nicotiana loci are coloured according to hummingbird colour categories(Supplementary Data Fig S3B) but are labelled with the accession name if the category includes only one taxon See Supplementary Data Video for an animation
of this graph
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1125
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
is the exception because it falls into an unexpected humming-bird colour category given the progenitor species (Fig 6)Section Repandae polyploids seem to have evolved to be eitherlike their maternal (Nicotiana nesophila N repanda andN stocktonii) or paternal (N nudicaulis) progenitor (Fig 6)The maternal progenitor N sylvestris is no longer sympatricwith any of the section Repandae polyploids therefore Nnesophila N repanda and N stocktonii can occupy the samepollination niche as their maternal progenitor without competi-tion Similarly section Suaveolentes is native to Australasiaexcept for one species in Namibia Africa and is not sympatricwith its progenitor sections in South America (Goodspeed1954) these polyploids and their diploid progenitors displaysimilar floral colours except N pauciflora which evolvedspectrally UVndashwhite flowers after the formation of sectionSuaveolentes (Fig 6) It is possible that there is less competitionfor pollinators and therefore reduced selective pressure to-wards floral colour diversification when polyploid species arenot sympatric with their diploid progenitors as is seen in
Iochrominae (Solanaceae) which have a broader range of floralcolours when species are sympatric (Muchhala et al 2014)However floral colour evolution can also be driven by geneticdrift or selection on pleiotropic effects of floral genes (Chittkaet al 2001 Rausher 2008) Furthermore anthocyanins andflavonoids are important for UV protection can defend againstfungi act as signal molecules and play a role in male fertility insome species in addition to their roles in signalling to pollina-tors (Shirley 1996)Over half of polyploids have an unexpected inheritance pat-
tern for chlorophyll if it is assumed that the presence of chloro-phyll is a dominant character Those polyploids that deviatefrom expectation span an age range of synthetic to 10myoand always lack chlorophyll in their petals which is in linewith the direction of the shifts observed in the divergence ofdiploid species (Supplementary Data Fig S6) In carnation thedifference in chlorophyll concentration between white andgreen flowers is likely caused by downregulation of chlorophyllbiosynthesis genes in white flowers genes involved in
TABLE 1 Polyploid and homoploid hybrid observed and expected floral colours
Species Spectral Bee Hummingbird Chlorophyll
Observed Expected Observed Expected Observed Expected Observed Expected
N tabacum 095-55 R W P UV-B B-G sUV-P W P N CN tabacum 51789 P W P B-G B-G P W P N CN tabacum lsquoChulumanirsquo W W P B-G B-G P W P N Csyn N tabacum QM P W P B B-G LP W P N Csyn N tabacum TH37 P W P B-G B-G P W P N CTH32 P W P B-G B-G P W P N CN rustica var asiatica G G LG G B-G G G W C CN rustica var pavonii G G G G B-G G G W C Csyn UP (homoploid) G G UV-G G B-G UV-G G W C Csyn F1 (homoploid) G G LG G B-G G G W C Csyn N rustica S0 G G LG G B-G G G W C Csyn N rustica S1 G G LG G B-G G G W C CN arentsii W G W B-G B-G P W C CN clevelandii W UV-W W B-G UV B-G P UV-W W N CN quadrivalvis TW18 W UV-W W B-G UV B-G W UV-W W C CN quadrivalvis 9047 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 1 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 2 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 5 W UV-W W B-G UV B-G W UV-W W C CN repanda W W UV-W B-G B-G UV W W UV-W N CN nesophila W W UV-W B-G B-G UV W W UV-W N CN stocktonii W W UV-W B-G B-G UV W W UV-W C CN nudicaulis UV-W W UV-W UV B-G UV UV-W W UV-W C CN benthamiana W W ndash ndash ndash ndash N CN forsteri W W ndash ndash ndash ndash C CN gossei W W ndash ndash ndash ndash N CN megalosiphon W W ndash ndash ndash ndash N CN occidentalis W W ndash ndash ndash ndash N CN suaveolens W W ndash ndash ndash ndash N C
N glauca 51725 Y W ndash ndash ndash ndash C CN glauca 51751 YG W ndash ndash ndash ndash C CN linearis TW77 UV-W W ndash ndash ndash ndash C CN linearis 9647 UV-W W ndash ndash ndash ndash C CN glutinosa P P W G ndash ndash ndash ndash C C
The top block consists of polyploid accessions and the bottom block includes homoploid hybridsItalic denotes an unexpected phenotype given the colour categories of the progenitorsProgenitor bee and hummingbird colour categories are unknown for section Suaveolentes and natural homoploid hybrids (see text)R red W white P pink G green UV-W UVndashwhite Y yellow UV-B UVndashblue B-G bluendashgreen B blue LG light green UV high UV UV-G
UVndashgreen sUV-P saturated UVndashpink LP light pink N no chlorophyll C chlorophyll syn synthetic 9047 represents 904750042 9647 represents964750099
1126 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
rewards (Raine et al 2006) Hummingbirds appear to have noinnate preferences for particular colours but are likewise goodat forming associations between flower visual displays and re-wards (Goldsmith and Goldsmith 1979 Chittka and Waser1997) Hummingbirds have red receptors whereas many insectsdo not Consequently red flowers are poorly detectable by beepollinators but conspicuous for hummingbirds Thereforehummingbird-visited flowers are often red whereas those polli-nated by bees typically have a range of other colours(Rodriguez-Girones and Santamaria 2004 Shrestha et al2013) Flowers visited by nocturnal pollinators are more oftenwhite than those pollinated in full daylight probably to maxi-mize their detectability in dim light conditions (Kevan et al1996) Because of such differences in affinities of various polli-nator classes to certain flower colours differences in flowercolour can contribute to restricting gene flow between pheno-types although they will rarely result in complete reproductiveisolation for this differences in multiple traits are typically es-sential In Aquilegia (Ranunculaceae) blue- red- and whiteyellow-flowered species are primarily pollinated by bees hum-mingbirds and hawkmoths respectively (Grant 1952 Whittalland Hodges 2007) In Petunia axillaris (Solanaceae) hawk-moths prefer white flowers to pink flowers transformed to ex-press ANTHOCYANIN2 whereas bumblebees prefer pinkANTHOCYANIN2 flowers to white flowers demonstrating thatexpression of a single gene can cause differences in pollinatorvisitation (Hoballah et al 2007) Similarly manipulation of asingle locus controlling carotenoid production inMimulus flow-ers (Phrymaceae) results in a pollinator shift reaffirming theimportance of floral colour in determining pollinator behaviour(Bradshaw and Schemske 2003)To analyse floral colour in the context of pollination it is
necessary to consider both colour theory and pollinator visualsystems There are several important differences between thecolour vision systems of humans and those of various pollinatortypes Humans and many insects are trichromatic havingthree discrete photoreceptor types however humans possessred- (with peak sensitivity (kmax) near 560 nm) green-(kmaxfrac14 535 nm) and blue-sensitive (kmaxfrac14 420 nm) photore-ceptors (Bowmaker and Dartnall 1980) whereas many insectshave ultraviolet- (UV- kmax 350 nm) blue- (kmax 440 nm)and green-sensitive (kmax 530 nm) receptors (Peitsch et al1992 Briscoe and Chittka 2001 Kelber et al 2003) Manybirds (Bowmaker 1998) and some butterflies (Kelber 2001)have tetrachromatic colour vision In birds photoreceptors aresensitive to red green blue and either violet or UV wave-lengths (Hart and Hunt 2007) Hummingbirds have four singlecone types with peak sensitivities in the UV (kmaxfrac14 370 nm)blue (kmaxfrac14 440 nm) bluendashgreen (kmaxfrac14 508 nm) and yellow(kmaxfrac14 560 nm) the sensitivity of the last extends significantlyinto the red (Herrera et al 2008) Hummingbirds can learn todistinguish near-UV light (370 nm) from darkness confirmingthat they use their UV receptors for colour vision at a behaviou-ral level (Goldsmith 1980) We will take into account these dif-ferences in pollinator perception as we examine floral colour inthe genus Nicotiana (Solanaceae)We investigate the evolution of floral colour across
Nicotiana (Solanaceae) in the context of polyploidy and hybrid-ization Nicotiana is an excellent group in which to study the ef-fects of hybridization as nearly half of the 76 species are
allotetraploids of different ages (Chase et al 2003Clarkson et al 2004 2005 Leitch et al 2008 Kelly et al2013) and several putative homoploid (diploid) hybrids havealso been detected (Clarkson et al 2010 Kelly et al 2010)which add to the reticulate nature of the genus These phyloge-netic studies have also been used to predict the closest livingdescendent species of the parents that formed the homoploidhybrid and allopolyploid species hereafter called progenitorspecies as shown in Fig 1 Some synthetic polyploidsmade from these progenitor species are also availableproviding insight into the immediate effects of polyploidy andhybridization We compare floral colours of Nicotianapolyploid and homoploid hybrids with those of their diploidprogenitorsBecause various animal groups have different sensitivities
to colour it is necessary to model colour perception ofspecific pollinator classes to understand the significance offloral colour signals Here we consider floral colours from abee perspective (Chittka 1992) which can also be used to rep-resent other trichromatic insects such as hawkmoths due tosimilarities in photoreceptor sensitivities (Kelber et al 2003)and a hummingbird perspective (Herrera et al 2008 Restrepo2013) Hummingbirds and hawkmoths are known tovisit Nicotiana species (Aigner and Scott 2002 Kaczorowskiet al 2005 Kessler and Baldwin 2006 Nattero and Cocucci2007)Pigments typically determine floral colour however few
studies have examined the specific pigments present inNicotiana petals Aharoni et al (2001) confirm the presence ofanthocyanin pigmentation in N tabacum which seems to bepredominantly cyanidin derivatives Spectral colour shifts canoccur in anthocyanins due to hydroxylation and methylationwhich result in different types of anthocyanins (Castaneda-Ovando et al 2009 Andersen and Jordheim 2010) and differ-ences in pH as well as copigmentation with other pigmentsincluding carotenoids and colourless flavonoids or metal ionscan also cause spectral shifts in the same anthocyanincompound (Grotewold 2006 Andersen and Jordheim 2010)The yellow flower colour of Nicotiana glauca is due to caroten-oid pigmentation (Zhu et al 2007) Crossing experimentsbetween diploid Nicotiana species suggest that the presence ofchlorophyll in corolla tissue is dominant (Brieger 1935)and similar results corroborate this in the carnation Dianthuscaryophyllus (Caryophyllaceae Ohmiya et al 2014)In this paper we seek to determine what types of spectral
reflectance are found within Nicotiana and how they appear tobee and hummingbird pollinators We focus on the conse-quences of polyploidy and interspecific hybridization on floralcolour evolution Specifically we aimed to test the hypothesesthat (1) polyploid and homoploid hybrids will have floralcolours that will resemble at least one of their progenitors indiscrete spectral bee and hummingbird floral colour categoriesobtained from cluster analyses (2) polyploid and homoploidhybrids will be positive for chlorophyll pigmentation in corollatissue if at least one progenitor has chlorophyll present in itspetals (due to evidence of the dominance of chlorophyll pig-mentation) (3) increased cell size potentially associated withpolyploidy affects the concentration of pigments and in turncolour intensity and (4) floral colour evolution is constrainedby phylogeny
1118 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
N tabacum 095-55
N tabacum 51789
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N sylvestris N tomentosiformis
N tabacum lt0middot2 myo
times
Natural Synthetic
A
timesN sylvestris N otophora TH32
TH32 synthetic polyploid
Synthetic
B
timesN rustica
var asiaticaN rustica
var pavoniisynthetic
UxPsynthetic PUE1 F1
synthetic N rusticaPUE1-R10
S0
synthetic N rustica
PUE1-R1 S1
N paniculata N undulata
N rustica lt0middot2 myo
Natural
Synthetic
C
timesN undulata N wigandioides N arentsii
N arentsii lt0middot2 myo
Natural
D
Natural Synthetic
timesN clevelandii N quadrivalvis
904750042N quadrivalvis
TW18N obtusifolia
lsquoBaldwinrsquo
N attenuata
lsquoBaldwinrsquo
N times obtusiata line 1
N times obtusiata line 2
N times obtusiata line 5
Section Polydicliae ~1 myoE
timesN repanda N nesophila N stocktonii N nudicaulisN sylvestris N obtusifolia
TW143
Section Repandae~4middot5 myoF
Natural
NoctifloraendashPetunioides homoploid hybrids
N noctiflora N petunioides N acuminata N attenuata N miersii N pauciflora N linearis TW77
N glauca 51725
N glauca 51751
times
Section PetunioidesSection Noctiflorae
H
N glutinosa
N otophora N setchellii N tomentosiformis N undulata N wigandioides N glutinosa
times
Section Tomentosae Section Undulatae
I
Section Suaveolentes ~10 myo
N noctiflora N petunioides N acuminata N attenuata N miersii N pauciflora
times
G
Section PetunioidesSection Noctiflorae
times
N sylvestris
N benthamiana
N forsteri
N gossei
N occidentalis
var hesperis
N megalosiphonN suaveolens
FIG 1 Floral colour as perceived by humans of polyploid and homoploid hybrid Nicotiana and their diploid progenitors Polyploid ages were estimated using a mo-lecular clock calibrated with the geological age of volcanic islands with endemic Nicotiana species (Clarkson et al 2005) Absolute dates (millions of years myo)estimated by the clock should be treated with caution however relative ages of different polyploid sections should reflect the true sequence of polyploidizationevents (A) Natural and synthetic polyploids of N tabacum (B) Synthetic polyploid TH32 (C) Natural and synthetic N rustica polyploids Synthetic hybrids includea homoploid from a reciprocal cross and a polyploid series (F1 homoploid and S0 and S1 polyploids) of the same parentage as natural N rustica (D) Nicotiana are-ntsii (E) Natural polyploids of section Polydicliae Synthetic N obtusiata polyploid lines were made from a cross between the N obtusifolia and N attenuata ac-cessions studied here (F) Section Repandae (G) Section Suaveolentes contains 26 polyploid species (six included in this study) Biogeographical analyses suggestthat section Suaveolentes originated15 million years ago (mya) before the aridification of Australia (Ladiges et al 2011) and this seems to be relatively congru-ent with the molecular clock results which place its origin at 10mya (H) Homoploid hybrids N glauca and N linearis (I) Homoploid hybrid N glutinosa
Photographs are scaled to the same size
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1119
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
MATERIALS AND METHODS
Petal cell area measurements
To assess whether an increase in ploidy results in larger petalcells cell area was measured from a subset of polyploids andtheir progenitors For Nicotiana sylvestris A04750326Nicotiana rustica var asiatica Nicotiana rustica var pavoniiNicotiana paniculata Nicotiana undulata and Nicotiana nudi-caulis mature flowers were taken from plants and the adaxialpetal surface was imprinted in Elite HD vinylpolysiloxane im-pression material (dental wax supplied by ZhermackHarrogate UK) The wax was left to set and then used as amould for making epoxy petal casts Devcon high-strength ep-oxy was mixed according to the manufacturerrsquos instructionspoured into the mould and allowed to set for 12 h The epoxyrelief was removed and coated with gold using a QuorumK756X sputter coater The samples were then imaged using aFEI Philips XL30 FEGSEM scanning electron microscope ForNicotiana obtusifolia var obtusifolia TW143 Nicotianarepanda and Nicotiana stocktonii only fixed material wasavailable whole mature flowers were fixed in formalinndashaceticacidndashalcohol (FAA) (60 ethanol 6 formaldehyde 5 acetic acid) for 72h before being transferred to a 70 ethanol(EtOH) wash for 24 h The samples were then dehydratedthrough an ethanol series of 2 h each in 70 80 and 90 andtwo washes in 100 EtOH The samples were dissected andthen dried in an Autosamdri 815B critical point dryer Thesesamples were sputter-coated and imaged as described aboveFor all samples images were taken mid-petal from an angleperpendicular to the surface to minimize parallax error Cellsize measurements were carried out in ImageJ (httpimagejnihgovij) The circumference of the cell base was drawn free-hand and area was calculated for 100ndash150 cells until the cu-mulative mean stabilized One-way ANOVA and Tukeyrsquoshonest significance tests were performed in RStudio version098490 (httpwwwrstudioorg) to compare cell area of poly-ploids with those of their progenitors repeating the tests foreach polyploid section
Spectral reflectance measurements
Spectral reflectance measurements were recorded for 60Nicotiana accessions (41 taxa Supplementary Data Table S1)three flowers from different plants where possible were usedfor each accession Reflectance spectra from three Nicotianaotophora accessions were pooled because the spectra weresimilarSpectral reflectance of flowers at anthesis was measured
from 300 to 700 nm using an Avantes AvaSpec-2048 spectro-photometer with an Avantes AvaLight-DHS light source andcalibrated with a barium sulphate white standard from lab-sphereVR Nicotiana mutabilis was also measured later as flow-ers change from white to pink when mature pink flowers areless likely to have a nectar reward but add to the attraction ofthe overall floral display and therefore are still relevant to pol-linators (R Kaczorowski University of Haifa Israel perscomm) Reflectance spectra express the proportion of lightreflected by the flower at any given wavelength Spectra werevisualized and exported in increments of 1 nm using the
program AvaSoft version 703 Full (Avantes BV Eerbeek TheNetherlands) and imported into ExcelSpectra for each accession or colour morph were averaged
and then smoothed three times using a rolling average over9 nm Spectra for all accessions were submitted to the FloralReflectance Database (FReD wwwreflectancecouk Arnoldet al 2010)Some spectra had a spike at 656 nm which corresponded
to a narrow peak in the light source spectrum suggesting thatthe spectra were saturated at 656 nm however smoothingserved to neutralize this spike Several spectra (Nicotiana are-ntsii N mutabilis Nicotiana suaveolens and Nicotiana wigan-dioides) included an anomalous reflectance minimum from 475to 500 nm which could not be explained by the light sourcespectrum Remeasured spectra of N arentsii N suaveolens andN wigandioides lacked this minimum but further material ofN mutabilis was unavailable so these spectra were includeddespite the anomalies
Calculation of colour loci in the bee colour hexagon
A reflectance spectrum can be represented as a single pointin the bee colour hexagon space (a graphical representation of abeersquos colour visual experience) based on the relative excitationof UV- blue- and green-sensitive photoreceptor types (Chittka1992) Vertices of this hexagon represent theoretical stateswhere one or two photoreceptor types are at maximal excitationwhereas at least one receptor type is at zero excitation (eg thetop vertex of the hexagon corresponds to maximal blue receptorexcitation and zero signal from UV and green receptorswhereas the top right vertex corresponds to maximal signal inboth blue and green receptors but no signal in the UV receptorand so forth see Supplementary Data Fig S1) The centre ororigin of the hexagon is achromatic Hue corresponds to angu-lar position around the origin whereas spectral purity or satura-tion increases with distance from the originBee colour hexagon coordinates were calculated for all
Nicotiana spectra Illumination was assumed to be sunlight(D65 Wyszecki and Stiles 1982) the background was repre-sented by an average leaf spectrum (Gumbert et al 1999)Honeybee photoreceptor spectral sensitivity functions wereused to determine bee colour hexagon coordinates these aresimilar to bumblebee and hawkmoth photoreceptor sensitivityfunctions so the bee colour hexagon can be used to approxi-mate the colour vision of these insects as well (Menzel et al1986 Peitsch et al 1992 Briscoe and Chittka 2001 Kelberet al 2003 and references therein Skorupski et al 2007) Theequations used to determine colour hexagon coordinates are asfollows where EG EB and EUV represent the excitation of thegreen blue and UV bee photoreceptors respectively elicitedby a spectrum (Chittka 1992)
x frac14ffiffiffi
3p
=2ethEG EUVTHORNy frac14 EB 05ethEUV thorn EGTHORN
Because the colour loci of Nicotiana flowers were mostlyclose to the centre of the colour space all colour hexagon dis-plays presented are scaled so that only the central 40 is
1120 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
shown the outline is therefore drawn as a dashed line This re-sults in a clearer spread of the colour loci to facilitate visual in-spection For a diagram explaining the colour hexagon seeSupplementary Data Fig S1
Calculation of colour loci in hummingbird colour space
For tetrachromatic hummingbirds we chose to model flowercolours in a 3-D colour opponent space because n ndash 1 colouropponent dimensions are necessary to code the informationfrom n colour receptors (Chittka 1996) The hummingbirdcolour space can be displayed as a rhombic dodecahedron with14 vertices (Restrepo 2013) Like the bee colour hexagonvertices of the space represent states where one two or threephotoreceptor types are at maximal excitation and at least onereceptor type is at zero excitation
Hummingbird colour space coordinates were calculated forall Nicotiana spectra Illumination was again assumed to besunlight (D65 Wyszecki and Stiles 1982) and the backgroundan average leaf spectrum (Gumbert et al 1999) as was usedfor the bee colour hexagon Photoreceptor spectral sensitivityfunctions from the green-backed firecrown hummingbird(Sephanoides sephanoides Herrera et al 2008) were used todetermine hummingbird colour space coordinates usingthe following equations (Restrepo 2013) where ER EGEB and EUV represent excitation of red green blue andUV hummingbird photoreceptors respectively elicited by aspectrum
x frac14ffiffiffiffiffiffiffiffi
3=4p
EB ffiffiffiffiffiffiffiffiffiffi
1=12p
ethEUV thorn EG thorn ERTHORN
y frac14ffiffiffiffiffiffiffiffi
2=3p
EG ffiffiffiffiffiffiffiffi
1=6p
ethEUV thorn ERTHORN
z frac14ffiffiffiffiffiffiffiffi
1=2p
ethEUV ERTHORN
RStudio was used to make 3D plots of the hummingbird col-our space and ImageJ version 148 (httpimagejnihgovij)was used to create an animation of the Nicotiana flower loci inthe hummingbird colour space Again Nicotiana flower colourloci are close to the origin in the hummingbird colour space sothe graphs presented display only the central portion (either 25or 50 ) of the colour space for clarity To further facilitateinterpretation of these graphs vertices representing individualexcitation of the red green blue and UV photoreceptor typesas well as their excitation vectors from the origin are shownin red green blue and black respectively Other vertices(representing excitation of two or three photoreceptor types) areshown in grey
Cluster analyses
Cluster analyses were used to group spectra based on spectralshape (corresponding to pigmentation) and their position inboth bee and hummingbird colour spaces For spectral colourcategories spectra were normalized to the same integral underthe curve in order to compare combinations of pigments notthe concentration of pigments A distance matrix was calculatedfrom the normalized spectral data in R version 302 (http
wwwR-projectorg) using the dist() function For the bee andhummingbird colour categories the input data were the (x y) or(x y z) coordinates of the spectra in the bee and hummingbirdcolour spaces respectivelyData were first imported into R The function hclust() was
used to perform agglomerative hierarchical clustering based onthe average pairwise distances between groups With this algo-rithm the observed points which are initially all deemed to bedistinct are iteratively assigned to groups until eventually allpoints belong to the same group At each step the average dis-tance between all groups (ie the mean distance from all pointsin group A to all points in group B if either one of these is asingle point then no averaging is needed) is calculated and thetwo groups with the smallest average distance are merged Theorder in which groups are merged can be used to construct adendrogram showing the spatial relationship between all datapoints We can also look at the distribution of merge distancesat each step in the algorithm and use this distribution to esti-mate how many groups are present in the data Points at whichthere is a steep increase in the average between-group distance(lsquoelbowrsquo points) highlight the spatial scale(s) at which there isclustering present in the data By using one of these elbowpoints as a cutoff in the algorithm we can arrive at a distancegrouping that captures the spatial clustering It should be notedthat the determination of where to draw the threshold in a clus-ter analysis is arbitrary but the use of one of these elbow pointsdoes yield meaningful clusters The determination of the spe-cific point from the elbow region to be used to define clusterswas further informed by visual inspection of reflectance spec-tra as well as distributions of colour loci in the perceptual col-our spaces It should be noted that the dendrograms relate tosimilarities in spectral reflectance as well as colour relation-ships perceived by bees and hummingbirds they do not showphylogenetic relationships
Ancestral state reconstruction
To examine the evolution of colour within a phylogeneticcontext ancestral state reconstructions were performed on treesinferred from plastid sequence data Only species for which flo-ral character data are available were included in these analysesBecause polyploid and homoploid hybrids originate via reticu-late evolutionary processes and therefore lack a history oftree-like evolution ancestral characters were reconstructed us-ing only non-hybrid diploid species The states observed inhybrid species were then compared with the ancestral state re-constructions Since sections Repandae and Suaveolentes havediversified to form several species following polyploidizationcharacters were reconstructed for these sections separately toexamine colour evolution subsequent to their origin Fornon-hybrid diploid species individual gene trees yieldsome conflicting topologies nevertheless key nodes for thepurposes of interpreting character evolution in hybrids arerecovered in multiple gene trees and are supported bysupernetwork analyses (Kelly et al 2010) Therefore plastiddata from previously published studies are suitable for theseanalysesPreviously published sequences (Clarkson et al 2004) from
four plastid regions (matK ndhF trnL-F and trnS-G) were
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1121
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
aligned separately using PRANKthornF (Loytynoja and Goldman2008) and then concatenated to create a combined plastid data-set before further optimization by eye in Mesquite version 274(Maddison and Maddison 2008) For Nicotiana attenuata weused GenBank accessions AB040009 and AY098697 for thematK and trnL-F regions respectively (due to likely misidenti-fication of N attenuata material used in Clarkson et al 2004see Clarkson et al 2010) the other two regions were scored asmissing data Phylogenetic reconstruction by Bayesian infer-ence was performed as described in Kelly et al (2013) with theexception that BayesTrees v13 (wwwevolutionreadingacukBayesTreeshtml) was used to construct 95 majority ruleconsensus trees For sections Repandae and Suaveolentes se-quences representing their putative maternal progenitors wereincluded during Bayesian inference to allow rooting of trees butwere pruned from the trees prior to ancestral statereconstructionAncestral states for spectral reflectance colour categories and
presenceabsence of chlorophyll in petals (data inSupplementary Data Table S2) were reconstructed using theparsimony reconstruction method in Mesquite version 274 un-der the unordered states assumption To account for topologicaluncertainty character states were reconstructed over all 36 000post-burn-in trees using the Trace Character Over Trees optionand summarized on the 95 majority rule consensus tree fromthe Bayesian analysis Ancestral states were not calculated forbee or hummingbird colour categories because these are per-ceptual systems and the same colour category can result fromdifferent combinations of pigments (eg both human pink andhuman white flowers which are positive and negative respec-tively for anthocyanin pigmentation are both classified as beebluendashgreen) thus a single colour category does not necessarilyhave a shared evolutionary history
Estimating expected polyploid and homoploid hybrid floralcolour
Polyploid and homoploid hybrid floral colours for each ac-cession were compared with those of their diploid progenitorsfor the spectral bee and hummingbird colour categories definedby cluster analyses Floral colour was classified as lsquoexpectedrsquo ifit fell in the colour category of at least one progenitor or lsquounex-pectedrsquo if it was different from both progenitors Polyploid andhomoploid hybrids were also compared with their diploid pro-genitors for the presence or absence of chlorophyll in corollatissue Chlorophyll absorbs at 675 nm in vivo (Haardt andMaske 1987) therefore the presence of chlorophyll can be in-ferred from reflectance spectra if there is a reflectance mini-mum at 675 nm The presence of chlorophyll in petal tissueappears to be dominant (Brieger 1935 Ohmiya et al 2014)Thus hybrids were classified as expected if they showed chlo-rophyll pigmentation and as unexpected if they did not becauseat least one progenitor possessed chlorophyll in all diploid pro-genitor combinations For natural homoploid hybrids and poly-ploid section Suaveolentes where progenitors can only bedefined to Nicotiana section level comparisons were madewith reconstructed ancestral characters thus only spectral col-our categories and the presenceabsence of chlorophyll wereexamined
Phylogenetic signal in floral traits
In order to statistically test for phylogenetic signal in the phe-notypic trait data (spectral reflectance bee and hummingbirdcolour perception) we used Mantel tests to examine the corre-lation between phylogenetic distance and each of the respectivecontinuous multidimensional traits (eg Cubo et al 2005Muchhala et al 2014) Analyses were restricted to diploid spe-cies excluding homoploid and polyploid hybrids Trees wereedited in Newick format to include additional tips with zerobranch lengths for taxa that are multiple in the trait datasetseither due to colour polymorphism (N otophora) or multipleaccessions (N sylvestris and N obtusifolia var obtusifolia)Statistical analyses were performed in R version 310
Phenotypic distance matrices were first calculated for the threetrait datasets using Euclidean distance and phylogenetic dis-tance matrices were calculated (1) as genetic distance from theplastid alignment and (2) for each of 36 000 post-burn-inBayesian trees using copheneticphylo() part of the APE pack-age version 31-2 (Paradis et al 2004) The second Bayesianset of tests was performed in order to account for evolutionaryprocesses such as saturation and to estimate how phylogeneticuncertainty affects the correlation Mantel tests were performedusing Pearsonrsquos product-moment correlation coefficient with10 000 permutations of each distance matrix to test for signifi-cance the mean P value and its standard deviation were calcu-lated for each set of 36 000 Mantel tests from the Bayesiantrees along with the percentage of trees that gave significantcorrelations The function mantel() from the vegan packagewas used (Oksanen et al 2013)
RESULTS
Petal cell area
Petal cell area was measured to determine whether an increasein ploidy results in larger floral cells Polyploid petal cell areawas significantly larger than in both progenitors in N rustica(ANOVA Ffrac14 371 dffrac14 3 Plt 2 10ndash16) accessions but wasintermediate between progenitors in section Repandae poly-ploids (ANOVA Ffrac14 2492 dffrac14 4 Plt 2 10ndash16 Fig 2)Accessions that were significantly different in cell area (withinpolyploid sections and their progenitors) are represented bydifferent letters above the bars in Fig 2 results from Tukeyrsquoshonest significance tests can be found in Supplementary DataTable S3
Cluster analyses
Nicotiana reflectance spectra were grouped into categoriesbased on spectral shape and position in the bee and humming-bird colour spaces using cluster analyses Bees and humming-birds have different photoreceptor sensitivities and we expectour cluster analyses to reflect these differences in sensoryequipment The analysis based on spectral shape yielded eightcolour categories which roughly corresponded to flowers per-ceived by human observers as magenta red pink UVndashwhitewhite yellow green and dark green (Fig 3) Nicotiana spectraare displayed by spectral colour category in Fig 4A BSupplementary Data Fig S2 The bee colour hexagon clustering
1122 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
resulted in 11 colour categories which fell into the followingareas of bee colour space saturated green UVndashblue high UVUVndashgreen green light green bluendashgreen dark green saturatedUVndashblue saturated UVndashgreen and blue (the last four categorieswere each represented by only a single accessionSupplementary Data Fig S3A) These groups are shown in thebee colour hexagon (Fig 4C) The hummingbird colour spacecluster analysis also produced 11 colour categories saturatedgreen green UVndashwhite UVndashgreen pink white UVndashpinkdark green light pink red and saturated UVndashpink (again thelast four categories include only a single accessionSupplementary Data Fig S3B) These groups are shown in thehummingbird colour space (Fig 4D) and the same graph isprovided as an animation to better display the 3D nature of thecolour space (Supplementary Data Video)
Evolution of spectral reflectance in polyploids and homoploidhybrids
To assess the evolution of polyploid floral colour polyploidand homoploid hybrid accessions were compared with those oftheir progenitors in spectral bee and hummingbird colour cate-gories as well as in the presenceabsence of chlorophyll Thediploid progenitors and approximate age of polyploids andhomoploid hybrids are found in Fig 1 and Supplementary DataTable S4 and the observed and expected floral colours of
polyploid and homoploid hybrids are found in Table 1 Mostpolyploid and homoploid hybrids were similar to at least oneprogenitor in spectral bee and hummingbird colour categoriesbut some fell into unexpected colour categories (Table 1Fig 5 Supplementary Data Fig S4 Supplementary Data FigS5) Over half of the polyploids unexpectedly lacked chloro-phyll (Table 1)
Evolution of colour characters in a phylogenetic context
Reconstructed character states are shown for spectral reflec-tance colour categories (Fig 6) and the presenceabsence ofchlorophyll in petals (Supplementary Data Fig S6) Bee andhummingbird colour categories are also shown for extant spe-cies on the plastid tree (Fig 6) Although the deepest nodeswere largely equivocal evolution of spectral reflectance colourin Nicotiana seemed to be dynamic (Fig 6) Green flowerslikely have three independent origins (1) in sectionsPaniculatae and Undulatae (2) in N langsdorffii and (3) inthe homoploid hybrid N glauca UVndashwhite flowers also seemto have arisen three times independently (1) in sectionTrigonophyllae (2) in N pauciflora and (3) in the homoploidhybrid N linearis Most polyploid and homoploid hybrid spe-cies exhibit a floral colour present in at least one of their pro-genitors However N tabacum 095-55 is red and N glauca isyellow and green unlike their progenitors UVndashwhite flowersseem to have evolved de novo in N linearis UVndashwhite flowersare also found in one of its progenitor sections (in N pauci-flora) but ancestral reconstructions indicate that the floral col-our was most likely white at the ancestral nodes within thesection (Fig 6) This suggests that the evolution of UVndashwhiteflowers in N pauciflora has occurred subsequent to the forma-tion of N linearis and that the two events are likely indepen-dent It is unclear whether UVndashwhite flowers also evolved denovo in N nudicaulis because the ancestral node of sectionRepandae is equivocal The presence of chlorophyll as inferredby light absorption at 675 nm (Haardt and Maske 1987) inNicotiana flowers is ancestral and has been lost three times inN sylvestris N noctiflora and the most recent common ances-tor of N acuminata and N pauciflora (Supplementary DataFig S6)Results from Mantel tests of phylogenetic signal for
Nicotiana floral traits for both genetic distance and the 36 000post-burn-in Bayesian trees are shown in Table 2 All floraltraits were significantly correlated with phylogenetic relation-ships for the Bayesian trees at a significance level of Plt 005Only spectral reflectance was significant for the genetic dis-tance tests whereas bee and hummingbird colour perceptionwere just above the Plt 005 threshold For the Bayesian trees901 662 and 932 of trees were significantly correlatedwith the spectral reflectance bee and hummingbird colour per-ception datasets respectively These results suggest that thesefloral traits are weakly constrained by phylogeny
DISCUSSION
Nicotiana is remarkable in its range of spectral reflectanceflower colours (white UVndashwhite pink magenta red yellowgreen and dark green Fig 3) and in the variety of pollinators
0
200
400
600
800
1000
1200C
ell
are
a (
microm
2)
N rustica
N paniculata
N undulata
N sylvestris
N obtusifolia
N nudicaulis
N repanda
N stocktonii
N rustica var asiatica
N rustica var pavonii
Repandae
a
a
b
b
cc
c c
d
Maternal progenitor
Paternal progenitor
Polyploids
Polyploids
Polyploids
FIG 2 Petal cell area from polyploids and their progenitors Within each poly-ploid group bars with different letters represent significantly different mean cell
areas
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1123
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
that visit the flowers (moth bird bee bat Knapp 2010) Theperception of these spectral colours also changes with visualsystem (bee or hummingbird) Here we describe a complex dy-namic in the evolution of floral colour in Nicotiana Spectral re-flectance and bee and hummingbird colour perception arecorrelated with phylogeny but multiple independent origins ofvarious combinations of pigmentation suggest that the evolutionof floral colour is not entirely phylogenetically constrained
Petal cell size evolution in polyploids
Cell size is expected to increase following polyploidizationdue to the increase in genome size (Beaulieu et al 2008) Thesignificantly larger petal cells of N rustica (lt02 million yearsold myo) and the intermediate cell size of section Repandae(45myo) suggest that polyploids may revert to a diploid-like cell size over time similar to the genome downsizing ob-served in polyploids (Leitch and Bennett 2004) Howeverpetal cell size differences within section Repandae do not seemto be linked to genome size N nudicaulis and N repanda sharesimilar cell sizes but have substantially different genome sizes(Leitch et al 2008) The concentration of pigment in floralcells is controlled by both the amount of pigment present andthe cell size Nicotiana rustica has significantly larger petalcells than its progenitors (close to the sum Fig 2) and displaysan intermediate brightness (the area under the reflectance curvea proxy for pigment concentration) between its progenitors
(Supplementary Data Fig S4D) This is expected if the poly-ploid inherits the sum of both cell size and amount of pigmentpresent from its progenitors
Polyploid divergence in floral colour
Many younger polyploids (lt02myo) display unexpectedfloral colours considering those of their diploid progenitorsNone of the natural and synthetic N tabacum accessions pos-sess chlorophyll which is unexpected given its presence in atleast one progenitor species Nicotiana tabacum 095-55 alsohas unexpected spectral bee and hummingbird colour giventhe colour categories of the progenitor species Similarly syn-thetic N tabacum QM has unexpected colours in bee and hum-mingbird perception Because this accession is synthetic theparents are known and thus its unexpected phenotype can beclassified as transgressive or outside the range of its progeni-tors due solely to polyploidy and hybridization Most N rusticaaccessions have unexpected bee colour (four of these aresynthetic and are therefore also transgressive) and N arentsiihas unexpected hummingbird colour (Table 1 Fig 6) Despitethe divergence of floral spectra associated with polyploidybehavioural studies are needed to determine whether the beeand hummingbird colour categories delineated here actuallyelicit different responses in pollinatorsMost older polyploids (1ndash10myo) are similar in floral col-
our category to at least one of their progenitors N clevelandii
synth
etic
N ru
stic
a P
UE
1-R
1 S
1
N ru
stic
a v
ar a
sia
tica
synth
etic
N ru
stic
a P
UE
1-R
10 S
0
synth
etic
PU
E1 F
1synth
etic
UxP
N b
enavid
esii
N ru
stic
a v
ar p
avonii
N p
anic
ula
taN
langsdorffii
N g
lauca 5
1751 g
reen
N u
ndula
taN
raim
ondii
N k
nig
htia
na
N g
lauca 5
1751 y
ello
w
N g
lauca 5
1725
N p
lum
bagin
ifolia
N g
ossei
N q
uadriv
alv
is T
W18
N o
ccid
enta
lis s
ubsp h
esperis
N
megalo
sip
hon
N s
uaveole
ns
N n
esophila
N s
tockto
nii
N q
uadriv
alv
is 9
04750042
N a
ttenuata
N re
panda
N s
ylv
estris
A04750326
N
benth
am
iana
N m
uta
bilis
CP
G3 w
hite
N
muta
bilis
CP
G12456 w
hite
N
wig
andio
ides
N a
cum
inata
N fo
rste
riN
cle
vela
ndii
N n
octiflo
raN
x o
btu
sia
ta lin
e 2
N
x o
btu
sia
ta lin
e 1
N
mie
rsii
N p
etu
nio
ides
N x
obtu
sia
ta lin
e 5
N
sylv
estris
6898
N
oto
phora
white
N
are
nts
iiT
H3
2N
tabacum
lsquoChulu
manirsquo
N o
btu
sifo
lia v
ar p
alm
eri
N o
btu
sifo
lia v
ar o
btu
sifo
lia lsquoB
ald
win
rsquoN
linearis
TW
77
N n
udic
aulis
N o
btu
sifo
lia v
ar o
btu
sifo
lia T
W143
N lin
earis
964750099
N
pauciflo
raN
tom
ento
sifo
rmis
N o
tophora
pin
k
N g
lutin
osa
synth
etic
N ta
bacum
TH
37
N
tabacum
51789
synth
etic
N ta
bacum
QM
N
tabacum
095-5
5
N s
etc
hellii
N m
uta
bilis
CP
G3 p
ink
N m
uta
bilis
CP
G12456 p
ink
10
8
6
4
2
0
Dis
tan
ce
Spectral categories
MagentaRedPinkUV-white
WhiteYellowDark greenGreen
Dark greenSaturated greenGreenUV-whiteUV-green
PinkLight pinkWhiteUV-pinkRedSaturated UV-pink
SpecBeeHum
SpectralDark greenSaturated greenSaturated UV-blueUV-blueSaturated UV-greenHigh UV
UV-greenGreenLight greenBlueBlue-green
Bee Hummingbird
FIG 3 Dendrograms based on distance cluster analyses for spectral reflectance Coloured circles on the dendrogram represent distinct colour categories as determinedby the chosen threshold (dashed line) Dendrograms were similarly constructed for bee and hummingbird colour (see Supplementary Data Fig S3) The lines of col-oured circles at the tips of the dendrograms signify the category each taxon is assigned to in spectral bee and hummingbird colour as labelled (Spec Bee and Hum)for comparison between spectral categories and those of different visual systems Diploids polyploids and homoploids are denoted by black blue and orange text
respectively
1124 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
0
0middot2
0middot4
0middot6
0middot8
1middot0
Reflecta
nce
Wavelength (nm)
0
0middot2
0middot4
0middot6
0middot8
1middot0
300 350 400 450 500 550 600 650 700 300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
N benavidesii
N glauca 51751 g
N langsdorffii
N paniculata
N raimondii
N rustica var asiatica
N rustica var pavonii
synthetic UxP
synthetic F1
syn N rustica S0
syn N rustica S1
N undulata
N glutinosa
N otophora p
N tabacum 51789
syn N tabacum QM
syn N tabacum TH37
N tomentosiformis
A B
G 50
R 50
B 50
UV 50
Hummingbird colour space
White
0middot40middot4
ndash0middot40middot4
0middot4
0middot4
y
z
UV-white
Pink
Synthetic N tabacum QM
UV-pink
N tabacum 095-55
N setchellii
Green
N langsdorffii
Saturated green
UV-green
C DBee colour hexagon
Blue-green
Light green
Green
Dark green
Sat green
UV-green
Sat UV-green
B
B-G
G
UV-G
UV
UV-B
1 acum2 aren3 atten4 bena5 benth6 clev7 forst8 glau259 glau51y10 glau51g11 glut12 goss13 knight14 langs15 lin964716 linTW7717 mega18 mier19 mut1w20 mut1p21 mut3w22mut3p23 neso24 noct25 nudi26 xobtus127 xobtus228 xobtus529 obtusB30 obtusTW31 obtuspalm
32 occhesp33 otoph w34 otoph p35 pani36 pauc37 petun38 plumba39 quad904740 quadTW1841 raim42 repa43 rustasi44 rustpav45 syn UxP46 syn F147 syn rust S048 syn rust S149 setch50 stock51 suav52 sylv689853 sylvA04754 tab0955555 tab5178956 tabchulu57 syntabQM58 syntabTH3759 tomtform60 undu61 wigan62TH32
High UV
UV-blue
Sat UV-blue
Blue
FIG 4 (A B) Nicotiana reflectance spectra from 300 to 700nm which roughly correspond to colours perceived by human observers as pink (A) and green (B) SeeSupplementary Data Fig S2 for other spectral colour categories Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploidhybrid taxa Abbreviations p pink syn synthetic g green (C) Colour hexagon displaying the distribution of Nicotiana colour loci in bee colour space The hexa-gon has been scaled so that vertices represent 40 excitation of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-GUVndashgreen Bee colour categories are delineated by coloured ovals sat saturated Nicotiana species abbreviations are as follows acum acuminata aren arentsiiatten attenuata benavid benavidesii benth benthamiana clev clevelandii forst forsteri glau25 glauca 51725 glau51y glauca 51751 yellow glau51g glauca51751 green glut glutinosa goss gossei knight knightiana langs langsdorffii lin9647 linearis 964750099 linTW77 linearis TW77 mega megalosiphon miermiersii mut1w mutabilis CPG12456 white mut1p mutabilis CPG12456 pink mut3w mutabilis CPG3 white mut3p mutabilis CPG3 pink neso nesophila noctnoctiflora nudi nudicaulis obtus1 obtusiata line 1 obtus2 obtusiata line 2 obtus5 obtusiata line 5 obtusB obtusifolia var obtusifolia lsquoBaldwinrsquoobtusTW obtusifolia var obtusifolia TW143 obtuspalm obtusifolia var palmeri occhesp occidentalis subsp hesperis otoph w otophora white otoph p otophorapink pani paniculata pauc pauciflora petun petunioides plumba plumbaginifolia quad9047 quadrivalvis 904750042 quadTW18 quadrivalvis TW18 raimraimondii repa repanda rustasi rustica var asiatica rustpav rustica var pavonii syn UP synthetic UP syn F1 synthetic PUE1 F1 synrust S0 synthetic rus-tica PUE1-R10 S0 synrust S1 synthetic rustica PUE1-R1 S1 setch setchellii stock stocktonii suav suaveolens sylv6898 sylvestris 6898 sylvA047 sylvestrisA04750326 tab09555 tabacum 095-55 tab51789 tabacum 51789 tabchulu tabacum lsquoChulumanirsquo syntabQM synthetic tabacum QM syntabTH37 synthetictabacum TH37 tomtform tomentosiformis undu undulata wigan wigandioides TH32 TH32 synthetic N sylvestrisN otophora polyploid (D) Distribution ofNicotiana spectral loci in hummingbird colour space Vertices of the hummingbird colour space represent 50 excitation of the photoreceptors single photoreceptortype vertices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows repre-sent the vectors of these photoreceptors from the origin of the hummingbird colour space Nicotiana loci are coloured according to hummingbird colour categories(Supplementary Data Fig S3B) but are labelled with the accession name if the category includes only one taxon See Supplementary Data Video for an animation
of this graph
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1125
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
is the exception because it falls into an unexpected humming-bird colour category given the progenitor species (Fig 6)Section Repandae polyploids seem to have evolved to be eitherlike their maternal (Nicotiana nesophila N repanda andN stocktonii) or paternal (N nudicaulis) progenitor (Fig 6)The maternal progenitor N sylvestris is no longer sympatricwith any of the section Repandae polyploids therefore Nnesophila N repanda and N stocktonii can occupy the samepollination niche as their maternal progenitor without competi-tion Similarly section Suaveolentes is native to Australasiaexcept for one species in Namibia Africa and is not sympatricwith its progenitor sections in South America (Goodspeed1954) these polyploids and their diploid progenitors displaysimilar floral colours except N pauciflora which evolvedspectrally UVndashwhite flowers after the formation of sectionSuaveolentes (Fig 6) It is possible that there is less competitionfor pollinators and therefore reduced selective pressure to-wards floral colour diversification when polyploid species arenot sympatric with their diploid progenitors as is seen in
Iochrominae (Solanaceae) which have a broader range of floralcolours when species are sympatric (Muchhala et al 2014)However floral colour evolution can also be driven by geneticdrift or selection on pleiotropic effects of floral genes (Chittkaet al 2001 Rausher 2008) Furthermore anthocyanins andflavonoids are important for UV protection can defend againstfungi act as signal molecules and play a role in male fertility insome species in addition to their roles in signalling to pollina-tors (Shirley 1996)Over half of polyploids have an unexpected inheritance pat-
tern for chlorophyll if it is assumed that the presence of chloro-phyll is a dominant character Those polyploids that deviatefrom expectation span an age range of synthetic to 10myoand always lack chlorophyll in their petals which is in linewith the direction of the shifts observed in the divergence ofdiploid species (Supplementary Data Fig S6) In carnation thedifference in chlorophyll concentration between white andgreen flowers is likely caused by downregulation of chlorophyllbiosynthesis genes in white flowers genes involved in
TABLE 1 Polyploid and homoploid hybrid observed and expected floral colours
Species Spectral Bee Hummingbird Chlorophyll
Observed Expected Observed Expected Observed Expected Observed Expected
N tabacum 095-55 R W P UV-B B-G sUV-P W P N CN tabacum 51789 P W P B-G B-G P W P N CN tabacum lsquoChulumanirsquo W W P B-G B-G P W P N Csyn N tabacum QM P W P B B-G LP W P N Csyn N tabacum TH37 P W P B-G B-G P W P N CTH32 P W P B-G B-G P W P N CN rustica var asiatica G G LG G B-G G G W C CN rustica var pavonii G G G G B-G G G W C Csyn UP (homoploid) G G UV-G G B-G UV-G G W C Csyn F1 (homoploid) G G LG G B-G G G W C Csyn N rustica S0 G G LG G B-G G G W C Csyn N rustica S1 G G LG G B-G G G W C CN arentsii W G W B-G B-G P W C CN clevelandii W UV-W W B-G UV B-G P UV-W W N CN quadrivalvis TW18 W UV-W W B-G UV B-G W UV-W W C CN quadrivalvis 9047 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 1 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 2 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 5 W UV-W W B-G UV B-G W UV-W W C CN repanda W W UV-W B-G B-G UV W W UV-W N CN nesophila W W UV-W B-G B-G UV W W UV-W N CN stocktonii W W UV-W B-G B-G UV W W UV-W C CN nudicaulis UV-W W UV-W UV B-G UV UV-W W UV-W C CN benthamiana W W ndash ndash ndash ndash N CN forsteri W W ndash ndash ndash ndash C CN gossei W W ndash ndash ndash ndash N CN megalosiphon W W ndash ndash ndash ndash N CN occidentalis W W ndash ndash ndash ndash N CN suaveolens W W ndash ndash ndash ndash N C
N glauca 51725 Y W ndash ndash ndash ndash C CN glauca 51751 YG W ndash ndash ndash ndash C CN linearis TW77 UV-W W ndash ndash ndash ndash C CN linearis 9647 UV-W W ndash ndash ndash ndash C CN glutinosa P P W G ndash ndash ndash ndash C C
The top block consists of polyploid accessions and the bottom block includes homoploid hybridsItalic denotes an unexpected phenotype given the colour categories of the progenitorsProgenitor bee and hummingbird colour categories are unknown for section Suaveolentes and natural homoploid hybrids (see text)R red W white P pink G green UV-W UVndashwhite Y yellow UV-B UVndashblue B-G bluendashgreen B blue LG light green UV high UV UV-G
UVndashgreen sUV-P saturated UVndashpink LP light pink N no chlorophyll C chlorophyll syn synthetic 9047 represents 904750042 9647 represents964750099
1126 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
N tabacum 095-55
N tabacum 51789
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N sylvestris N tomentosiformis
N tabacum lt0middot2 myo
times
Natural Synthetic
A
timesN sylvestris N otophora TH32
TH32 synthetic polyploid
Synthetic
B
timesN rustica
var asiaticaN rustica
var pavoniisynthetic
UxPsynthetic PUE1 F1
synthetic N rusticaPUE1-R10
S0
synthetic N rustica
PUE1-R1 S1
N paniculata N undulata
N rustica lt0middot2 myo
Natural
Synthetic
C
timesN undulata N wigandioides N arentsii
N arentsii lt0middot2 myo
Natural
D
Natural Synthetic
timesN clevelandii N quadrivalvis
904750042N quadrivalvis
TW18N obtusifolia
lsquoBaldwinrsquo
N attenuata
lsquoBaldwinrsquo
N times obtusiata line 1
N times obtusiata line 2
N times obtusiata line 5
Section Polydicliae ~1 myoE
timesN repanda N nesophila N stocktonii N nudicaulisN sylvestris N obtusifolia
TW143
Section Repandae~4middot5 myoF
Natural
NoctifloraendashPetunioides homoploid hybrids
N noctiflora N petunioides N acuminata N attenuata N miersii N pauciflora N linearis TW77
N glauca 51725
N glauca 51751
times
Section PetunioidesSection Noctiflorae
H
N glutinosa
N otophora N setchellii N tomentosiformis N undulata N wigandioides N glutinosa
times
Section Tomentosae Section Undulatae
I
Section Suaveolentes ~10 myo
N noctiflora N petunioides N acuminata N attenuata N miersii N pauciflora
times
G
Section PetunioidesSection Noctiflorae
times
N sylvestris
N benthamiana
N forsteri
N gossei
N occidentalis
var hesperis
N megalosiphonN suaveolens
FIG 1 Floral colour as perceived by humans of polyploid and homoploid hybrid Nicotiana and their diploid progenitors Polyploid ages were estimated using a mo-lecular clock calibrated with the geological age of volcanic islands with endemic Nicotiana species (Clarkson et al 2005) Absolute dates (millions of years myo)estimated by the clock should be treated with caution however relative ages of different polyploid sections should reflect the true sequence of polyploidizationevents (A) Natural and synthetic polyploids of N tabacum (B) Synthetic polyploid TH32 (C) Natural and synthetic N rustica polyploids Synthetic hybrids includea homoploid from a reciprocal cross and a polyploid series (F1 homoploid and S0 and S1 polyploids) of the same parentage as natural N rustica (D) Nicotiana are-ntsii (E) Natural polyploids of section Polydicliae Synthetic N obtusiata polyploid lines were made from a cross between the N obtusifolia and N attenuata ac-cessions studied here (F) Section Repandae (G) Section Suaveolentes contains 26 polyploid species (six included in this study) Biogeographical analyses suggestthat section Suaveolentes originated15 million years ago (mya) before the aridification of Australia (Ladiges et al 2011) and this seems to be relatively congru-ent with the molecular clock results which place its origin at 10mya (H) Homoploid hybrids N glauca and N linearis (I) Homoploid hybrid N glutinosa
Photographs are scaled to the same size
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1119
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
MATERIALS AND METHODS
Petal cell area measurements
To assess whether an increase in ploidy results in larger petalcells cell area was measured from a subset of polyploids andtheir progenitors For Nicotiana sylvestris A04750326Nicotiana rustica var asiatica Nicotiana rustica var pavoniiNicotiana paniculata Nicotiana undulata and Nicotiana nudi-caulis mature flowers were taken from plants and the adaxialpetal surface was imprinted in Elite HD vinylpolysiloxane im-pression material (dental wax supplied by ZhermackHarrogate UK) The wax was left to set and then used as amould for making epoxy petal casts Devcon high-strength ep-oxy was mixed according to the manufacturerrsquos instructionspoured into the mould and allowed to set for 12 h The epoxyrelief was removed and coated with gold using a QuorumK756X sputter coater The samples were then imaged using aFEI Philips XL30 FEGSEM scanning electron microscope ForNicotiana obtusifolia var obtusifolia TW143 Nicotianarepanda and Nicotiana stocktonii only fixed material wasavailable whole mature flowers were fixed in formalinndashaceticacidndashalcohol (FAA) (60 ethanol 6 formaldehyde 5 acetic acid) for 72h before being transferred to a 70 ethanol(EtOH) wash for 24 h The samples were then dehydratedthrough an ethanol series of 2 h each in 70 80 and 90 andtwo washes in 100 EtOH The samples were dissected andthen dried in an Autosamdri 815B critical point dryer Thesesamples were sputter-coated and imaged as described aboveFor all samples images were taken mid-petal from an angleperpendicular to the surface to minimize parallax error Cellsize measurements were carried out in ImageJ (httpimagejnihgovij) The circumference of the cell base was drawn free-hand and area was calculated for 100ndash150 cells until the cu-mulative mean stabilized One-way ANOVA and Tukeyrsquoshonest significance tests were performed in RStudio version098490 (httpwwwrstudioorg) to compare cell area of poly-ploids with those of their progenitors repeating the tests foreach polyploid section
Spectral reflectance measurements
Spectral reflectance measurements were recorded for 60Nicotiana accessions (41 taxa Supplementary Data Table S1)three flowers from different plants where possible were usedfor each accession Reflectance spectra from three Nicotianaotophora accessions were pooled because the spectra weresimilarSpectral reflectance of flowers at anthesis was measured
from 300 to 700 nm using an Avantes AvaSpec-2048 spectro-photometer with an Avantes AvaLight-DHS light source andcalibrated with a barium sulphate white standard from lab-sphereVR Nicotiana mutabilis was also measured later as flow-ers change from white to pink when mature pink flowers areless likely to have a nectar reward but add to the attraction ofthe overall floral display and therefore are still relevant to pol-linators (R Kaczorowski University of Haifa Israel perscomm) Reflectance spectra express the proportion of lightreflected by the flower at any given wavelength Spectra werevisualized and exported in increments of 1 nm using the
program AvaSoft version 703 Full (Avantes BV Eerbeek TheNetherlands) and imported into ExcelSpectra for each accession or colour morph were averaged
and then smoothed three times using a rolling average over9 nm Spectra for all accessions were submitted to the FloralReflectance Database (FReD wwwreflectancecouk Arnoldet al 2010)Some spectra had a spike at 656 nm which corresponded
to a narrow peak in the light source spectrum suggesting thatthe spectra were saturated at 656 nm however smoothingserved to neutralize this spike Several spectra (Nicotiana are-ntsii N mutabilis Nicotiana suaveolens and Nicotiana wigan-dioides) included an anomalous reflectance minimum from 475to 500 nm which could not be explained by the light sourcespectrum Remeasured spectra of N arentsii N suaveolens andN wigandioides lacked this minimum but further material ofN mutabilis was unavailable so these spectra were includeddespite the anomalies
Calculation of colour loci in the bee colour hexagon
A reflectance spectrum can be represented as a single pointin the bee colour hexagon space (a graphical representation of abeersquos colour visual experience) based on the relative excitationof UV- blue- and green-sensitive photoreceptor types (Chittka1992) Vertices of this hexagon represent theoretical stateswhere one or two photoreceptor types are at maximal excitationwhereas at least one receptor type is at zero excitation (eg thetop vertex of the hexagon corresponds to maximal blue receptorexcitation and zero signal from UV and green receptorswhereas the top right vertex corresponds to maximal signal inboth blue and green receptors but no signal in the UV receptorand so forth see Supplementary Data Fig S1) The centre ororigin of the hexagon is achromatic Hue corresponds to angu-lar position around the origin whereas spectral purity or satura-tion increases with distance from the originBee colour hexagon coordinates were calculated for all
Nicotiana spectra Illumination was assumed to be sunlight(D65 Wyszecki and Stiles 1982) the background was repre-sented by an average leaf spectrum (Gumbert et al 1999)Honeybee photoreceptor spectral sensitivity functions wereused to determine bee colour hexagon coordinates these aresimilar to bumblebee and hawkmoth photoreceptor sensitivityfunctions so the bee colour hexagon can be used to approxi-mate the colour vision of these insects as well (Menzel et al1986 Peitsch et al 1992 Briscoe and Chittka 2001 Kelberet al 2003 and references therein Skorupski et al 2007) Theequations used to determine colour hexagon coordinates are asfollows where EG EB and EUV represent the excitation of thegreen blue and UV bee photoreceptors respectively elicitedby a spectrum (Chittka 1992)
x frac14ffiffiffi
3p
=2ethEG EUVTHORNy frac14 EB 05ethEUV thorn EGTHORN
Because the colour loci of Nicotiana flowers were mostlyclose to the centre of the colour space all colour hexagon dis-plays presented are scaled so that only the central 40 is
1120 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
shown the outline is therefore drawn as a dashed line This re-sults in a clearer spread of the colour loci to facilitate visual in-spection For a diagram explaining the colour hexagon seeSupplementary Data Fig S1
Calculation of colour loci in hummingbird colour space
For tetrachromatic hummingbirds we chose to model flowercolours in a 3-D colour opponent space because n ndash 1 colouropponent dimensions are necessary to code the informationfrom n colour receptors (Chittka 1996) The hummingbirdcolour space can be displayed as a rhombic dodecahedron with14 vertices (Restrepo 2013) Like the bee colour hexagonvertices of the space represent states where one two or threephotoreceptor types are at maximal excitation and at least onereceptor type is at zero excitation
Hummingbird colour space coordinates were calculated forall Nicotiana spectra Illumination was again assumed to besunlight (D65 Wyszecki and Stiles 1982) and the backgroundan average leaf spectrum (Gumbert et al 1999) as was usedfor the bee colour hexagon Photoreceptor spectral sensitivityfunctions from the green-backed firecrown hummingbird(Sephanoides sephanoides Herrera et al 2008) were used todetermine hummingbird colour space coordinates usingthe following equations (Restrepo 2013) where ER EGEB and EUV represent excitation of red green blue andUV hummingbird photoreceptors respectively elicited by aspectrum
x frac14ffiffiffiffiffiffiffiffi
3=4p
EB ffiffiffiffiffiffiffiffiffiffi
1=12p
ethEUV thorn EG thorn ERTHORN
y frac14ffiffiffiffiffiffiffiffi
2=3p
EG ffiffiffiffiffiffiffiffi
1=6p
ethEUV thorn ERTHORN
z frac14ffiffiffiffiffiffiffiffi
1=2p
ethEUV ERTHORN
RStudio was used to make 3D plots of the hummingbird col-our space and ImageJ version 148 (httpimagejnihgovij)was used to create an animation of the Nicotiana flower loci inthe hummingbird colour space Again Nicotiana flower colourloci are close to the origin in the hummingbird colour space sothe graphs presented display only the central portion (either 25or 50 ) of the colour space for clarity To further facilitateinterpretation of these graphs vertices representing individualexcitation of the red green blue and UV photoreceptor typesas well as their excitation vectors from the origin are shownin red green blue and black respectively Other vertices(representing excitation of two or three photoreceptor types) areshown in grey
Cluster analyses
Cluster analyses were used to group spectra based on spectralshape (corresponding to pigmentation) and their position inboth bee and hummingbird colour spaces For spectral colourcategories spectra were normalized to the same integral underthe curve in order to compare combinations of pigments notthe concentration of pigments A distance matrix was calculatedfrom the normalized spectral data in R version 302 (http
wwwR-projectorg) using the dist() function For the bee andhummingbird colour categories the input data were the (x y) or(x y z) coordinates of the spectra in the bee and hummingbirdcolour spaces respectivelyData were first imported into R The function hclust() was
used to perform agglomerative hierarchical clustering based onthe average pairwise distances between groups With this algo-rithm the observed points which are initially all deemed to bedistinct are iteratively assigned to groups until eventually allpoints belong to the same group At each step the average dis-tance between all groups (ie the mean distance from all pointsin group A to all points in group B if either one of these is asingle point then no averaging is needed) is calculated and thetwo groups with the smallest average distance are merged Theorder in which groups are merged can be used to construct adendrogram showing the spatial relationship between all datapoints We can also look at the distribution of merge distancesat each step in the algorithm and use this distribution to esti-mate how many groups are present in the data Points at whichthere is a steep increase in the average between-group distance(lsquoelbowrsquo points) highlight the spatial scale(s) at which there isclustering present in the data By using one of these elbowpoints as a cutoff in the algorithm we can arrive at a distancegrouping that captures the spatial clustering It should be notedthat the determination of where to draw the threshold in a clus-ter analysis is arbitrary but the use of one of these elbow pointsdoes yield meaningful clusters The determination of the spe-cific point from the elbow region to be used to define clusterswas further informed by visual inspection of reflectance spec-tra as well as distributions of colour loci in the perceptual col-our spaces It should be noted that the dendrograms relate tosimilarities in spectral reflectance as well as colour relation-ships perceived by bees and hummingbirds they do not showphylogenetic relationships
Ancestral state reconstruction
To examine the evolution of colour within a phylogeneticcontext ancestral state reconstructions were performed on treesinferred from plastid sequence data Only species for which flo-ral character data are available were included in these analysesBecause polyploid and homoploid hybrids originate via reticu-late evolutionary processes and therefore lack a history oftree-like evolution ancestral characters were reconstructed us-ing only non-hybrid diploid species The states observed inhybrid species were then compared with the ancestral state re-constructions Since sections Repandae and Suaveolentes havediversified to form several species following polyploidizationcharacters were reconstructed for these sections separately toexamine colour evolution subsequent to their origin Fornon-hybrid diploid species individual gene trees yieldsome conflicting topologies nevertheless key nodes for thepurposes of interpreting character evolution in hybrids arerecovered in multiple gene trees and are supported bysupernetwork analyses (Kelly et al 2010) Therefore plastiddata from previously published studies are suitable for theseanalysesPreviously published sequences (Clarkson et al 2004) from
four plastid regions (matK ndhF trnL-F and trnS-G) were
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1121
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
aligned separately using PRANKthornF (Loytynoja and Goldman2008) and then concatenated to create a combined plastid data-set before further optimization by eye in Mesquite version 274(Maddison and Maddison 2008) For Nicotiana attenuata weused GenBank accessions AB040009 and AY098697 for thematK and trnL-F regions respectively (due to likely misidenti-fication of N attenuata material used in Clarkson et al 2004see Clarkson et al 2010) the other two regions were scored asmissing data Phylogenetic reconstruction by Bayesian infer-ence was performed as described in Kelly et al (2013) with theexception that BayesTrees v13 (wwwevolutionreadingacukBayesTreeshtml) was used to construct 95 majority ruleconsensus trees For sections Repandae and Suaveolentes se-quences representing their putative maternal progenitors wereincluded during Bayesian inference to allow rooting of trees butwere pruned from the trees prior to ancestral statereconstructionAncestral states for spectral reflectance colour categories and
presenceabsence of chlorophyll in petals (data inSupplementary Data Table S2) were reconstructed using theparsimony reconstruction method in Mesquite version 274 un-der the unordered states assumption To account for topologicaluncertainty character states were reconstructed over all 36 000post-burn-in trees using the Trace Character Over Trees optionand summarized on the 95 majority rule consensus tree fromthe Bayesian analysis Ancestral states were not calculated forbee or hummingbird colour categories because these are per-ceptual systems and the same colour category can result fromdifferent combinations of pigments (eg both human pink andhuman white flowers which are positive and negative respec-tively for anthocyanin pigmentation are both classified as beebluendashgreen) thus a single colour category does not necessarilyhave a shared evolutionary history
Estimating expected polyploid and homoploid hybrid floralcolour
Polyploid and homoploid hybrid floral colours for each ac-cession were compared with those of their diploid progenitorsfor the spectral bee and hummingbird colour categories definedby cluster analyses Floral colour was classified as lsquoexpectedrsquo ifit fell in the colour category of at least one progenitor or lsquounex-pectedrsquo if it was different from both progenitors Polyploid andhomoploid hybrids were also compared with their diploid pro-genitors for the presence or absence of chlorophyll in corollatissue Chlorophyll absorbs at 675 nm in vivo (Haardt andMaske 1987) therefore the presence of chlorophyll can be in-ferred from reflectance spectra if there is a reflectance mini-mum at 675 nm The presence of chlorophyll in petal tissueappears to be dominant (Brieger 1935 Ohmiya et al 2014)Thus hybrids were classified as expected if they showed chlo-rophyll pigmentation and as unexpected if they did not becauseat least one progenitor possessed chlorophyll in all diploid pro-genitor combinations For natural homoploid hybrids and poly-ploid section Suaveolentes where progenitors can only bedefined to Nicotiana section level comparisons were madewith reconstructed ancestral characters thus only spectral col-our categories and the presenceabsence of chlorophyll wereexamined
Phylogenetic signal in floral traits
In order to statistically test for phylogenetic signal in the phe-notypic trait data (spectral reflectance bee and hummingbirdcolour perception) we used Mantel tests to examine the corre-lation between phylogenetic distance and each of the respectivecontinuous multidimensional traits (eg Cubo et al 2005Muchhala et al 2014) Analyses were restricted to diploid spe-cies excluding homoploid and polyploid hybrids Trees wereedited in Newick format to include additional tips with zerobranch lengths for taxa that are multiple in the trait datasetseither due to colour polymorphism (N otophora) or multipleaccessions (N sylvestris and N obtusifolia var obtusifolia)Statistical analyses were performed in R version 310
Phenotypic distance matrices were first calculated for the threetrait datasets using Euclidean distance and phylogenetic dis-tance matrices were calculated (1) as genetic distance from theplastid alignment and (2) for each of 36 000 post-burn-inBayesian trees using copheneticphylo() part of the APE pack-age version 31-2 (Paradis et al 2004) The second Bayesianset of tests was performed in order to account for evolutionaryprocesses such as saturation and to estimate how phylogeneticuncertainty affects the correlation Mantel tests were performedusing Pearsonrsquos product-moment correlation coefficient with10 000 permutations of each distance matrix to test for signifi-cance the mean P value and its standard deviation were calcu-lated for each set of 36 000 Mantel tests from the Bayesiantrees along with the percentage of trees that gave significantcorrelations The function mantel() from the vegan packagewas used (Oksanen et al 2013)
RESULTS
Petal cell area
Petal cell area was measured to determine whether an increasein ploidy results in larger floral cells Polyploid petal cell areawas significantly larger than in both progenitors in N rustica(ANOVA Ffrac14 371 dffrac14 3 Plt 2 10ndash16) accessions but wasintermediate between progenitors in section Repandae poly-ploids (ANOVA Ffrac14 2492 dffrac14 4 Plt 2 10ndash16 Fig 2)Accessions that were significantly different in cell area (withinpolyploid sections and their progenitors) are represented bydifferent letters above the bars in Fig 2 results from Tukeyrsquoshonest significance tests can be found in Supplementary DataTable S3
Cluster analyses
Nicotiana reflectance spectra were grouped into categoriesbased on spectral shape and position in the bee and humming-bird colour spaces using cluster analyses Bees and humming-birds have different photoreceptor sensitivities and we expectour cluster analyses to reflect these differences in sensoryequipment The analysis based on spectral shape yielded eightcolour categories which roughly corresponded to flowers per-ceived by human observers as magenta red pink UVndashwhitewhite yellow green and dark green (Fig 3) Nicotiana spectraare displayed by spectral colour category in Fig 4A BSupplementary Data Fig S2 The bee colour hexagon clustering
1122 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
resulted in 11 colour categories which fell into the followingareas of bee colour space saturated green UVndashblue high UVUVndashgreen green light green bluendashgreen dark green saturatedUVndashblue saturated UVndashgreen and blue (the last four categorieswere each represented by only a single accessionSupplementary Data Fig S3A) These groups are shown in thebee colour hexagon (Fig 4C) The hummingbird colour spacecluster analysis also produced 11 colour categories saturatedgreen green UVndashwhite UVndashgreen pink white UVndashpinkdark green light pink red and saturated UVndashpink (again thelast four categories include only a single accessionSupplementary Data Fig S3B) These groups are shown in thehummingbird colour space (Fig 4D) and the same graph isprovided as an animation to better display the 3D nature of thecolour space (Supplementary Data Video)
Evolution of spectral reflectance in polyploids and homoploidhybrids
To assess the evolution of polyploid floral colour polyploidand homoploid hybrid accessions were compared with those oftheir progenitors in spectral bee and hummingbird colour cate-gories as well as in the presenceabsence of chlorophyll Thediploid progenitors and approximate age of polyploids andhomoploid hybrids are found in Fig 1 and Supplementary DataTable S4 and the observed and expected floral colours of
polyploid and homoploid hybrids are found in Table 1 Mostpolyploid and homoploid hybrids were similar to at least oneprogenitor in spectral bee and hummingbird colour categoriesbut some fell into unexpected colour categories (Table 1Fig 5 Supplementary Data Fig S4 Supplementary Data FigS5) Over half of the polyploids unexpectedly lacked chloro-phyll (Table 1)
Evolution of colour characters in a phylogenetic context
Reconstructed character states are shown for spectral reflec-tance colour categories (Fig 6) and the presenceabsence ofchlorophyll in petals (Supplementary Data Fig S6) Bee andhummingbird colour categories are also shown for extant spe-cies on the plastid tree (Fig 6) Although the deepest nodeswere largely equivocal evolution of spectral reflectance colourin Nicotiana seemed to be dynamic (Fig 6) Green flowerslikely have three independent origins (1) in sectionsPaniculatae and Undulatae (2) in N langsdorffii and (3) inthe homoploid hybrid N glauca UVndashwhite flowers also seemto have arisen three times independently (1) in sectionTrigonophyllae (2) in N pauciflora and (3) in the homoploidhybrid N linearis Most polyploid and homoploid hybrid spe-cies exhibit a floral colour present in at least one of their pro-genitors However N tabacum 095-55 is red and N glauca isyellow and green unlike their progenitors UVndashwhite flowersseem to have evolved de novo in N linearis UVndashwhite flowersare also found in one of its progenitor sections (in N pauci-flora) but ancestral reconstructions indicate that the floral col-our was most likely white at the ancestral nodes within thesection (Fig 6) This suggests that the evolution of UVndashwhiteflowers in N pauciflora has occurred subsequent to the forma-tion of N linearis and that the two events are likely indepen-dent It is unclear whether UVndashwhite flowers also evolved denovo in N nudicaulis because the ancestral node of sectionRepandae is equivocal The presence of chlorophyll as inferredby light absorption at 675 nm (Haardt and Maske 1987) inNicotiana flowers is ancestral and has been lost three times inN sylvestris N noctiflora and the most recent common ances-tor of N acuminata and N pauciflora (Supplementary DataFig S6)Results from Mantel tests of phylogenetic signal for
Nicotiana floral traits for both genetic distance and the 36 000post-burn-in Bayesian trees are shown in Table 2 All floraltraits were significantly correlated with phylogenetic relation-ships for the Bayesian trees at a significance level of Plt 005Only spectral reflectance was significant for the genetic dis-tance tests whereas bee and hummingbird colour perceptionwere just above the Plt 005 threshold For the Bayesian trees901 662 and 932 of trees were significantly correlatedwith the spectral reflectance bee and hummingbird colour per-ception datasets respectively These results suggest that thesefloral traits are weakly constrained by phylogeny
DISCUSSION
Nicotiana is remarkable in its range of spectral reflectanceflower colours (white UVndashwhite pink magenta red yellowgreen and dark green Fig 3) and in the variety of pollinators
0
200
400
600
800
1000
1200C
ell
are
a (
microm
2)
N rustica
N paniculata
N undulata
N sylvestris
N obtusifolia
N nudicaulis
N repanda
N stocktonii
N rustica var asiatica
N rustica var pavonii
Repandae
a
a
b
b
cc
c c
d
Maternal progenitor
Paternal progenitor
Polyploids
Polyploids
Polyploids
FIG 2 Petal cell area from polyploids and their progenitors Within each poly-ploid group bars with different letters represent significantly different mean cell
areas
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1123
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
that visit the flowers (moth bird bee bat Knapp 2010) Theperception of these spectral colours also changes with visualsystem (bee or hummingbird) Here we describe a complex dy-namic in the evolution of floral colour in Nicotiana Spectral re-flectance and bee and hummingbird colour perception arecorrelated with phylogeny but multiple independent origins ofvarious combinations of pigmentation suggest that the evolutionof floral colour is not entirely phylogenetically constrained
Petal cell size evolution in polyploids
Cell size is expected to increase following polyploidizationdue to the increase in genome size (Beaulieu et al 2008) Thesignificantly larger petal cells of N rustica (lt02 million yearsold myo) and the intermediate cell size of section Repandae(45myo) suggest that polyploids may revert to a diploid-like cell size over time similar to the genome downsizing ob-served in polyploids (Leitch and Bennett 2004) Howeverpetal cell size differences within section Repandae do not seemto be linked to genome size N nudicaulis and N repanda sharesimilar cell sizes but have substantially different genome sizes(Leitch et al 2008) The concentration of pigment in floralcells is controlled by both the amount of pigment present andthe cell size Nicotiana rustica has significantly larger petalcells than its progenitors (close to the sum Fig 2) and displaysan intermediate brightness (the area under the reflectance curvea proxy for pigment concentration) between its progenitors
(Supplementary Data Fig S4D) This is expected if the poly-ploid inherits the sum of both cell size and amount of pigmentpresent from its progenitors
Polyploid divergence in floral colour
Many younger polyploids (lt02myo) display unexpectedfloral colours considering those of their diploid progenitorsNone of the natural and synthetic N tabacum accessions pos-sess chlorophyll which is unexpected given its presence in atleast one progenitor species Nicotiana tabacum 095-55 alsohas unexpected spectral bee and hummingbird colour giventhe colour categories of the progenitor species Similarly syn-thetic N tabacum QM has unexpected colours in bee and hum-mingbird perception Because this accession is synthetic theparents are known and thus its unexpected phenotype can beclassified as transgressive or outside the range of its progeni-tors due solely to polyploidy and hybridization Most N rusticaaccessions have unexpected bee colour (four of these aresynthetic and are therefore also transgressive) and N arentsiihas unexpected hummingbird colour (Table 1 Fig 6) Despitethe divergence of floral spectra associated with polyploidybehavioural studies are needed to determine whether the beeand hummingbird colour categories delineated here actuallyelicit different responses in pollinatorsMost older polyploids (1ndash10myo) are similar in floral col-
our category to at least one of their progenitors N clevelandii
synth
etic
N ru
stic
a P
UE
1-R
1 S
1
N ru
stic
a v
ar a
sia
tica
synth
etic
N ru
stic
a P
UE
1-R
10 S
0
synth
etic
PU
E1 F
1synth
etic
UxP
N b
enavid
esii
N ru
stic
a v
ar p
avonii
N p
anic
ula
taN
langsdorffii
N g
lauca 5
1751 g
reen
N u
ndula
taN
raim
ondii
N k
nig
htia
na
N g
lauca 5
1751 y
ello
w
N g
lauca 5
1725
N p
lum
bagin
ifolia
N g
ossei
N q
uadriv
alv
is T
W18
N o
ccid
enta
lis s
ubsp h
esperis
N
megalo
sip
hon
N s
uaveole
ns
N n
esophila
N s
tockto
nii
N q
uadriv
alv
is 9
04750042
N a
ttenuata
N re
panda
N s
ylv
estris
A04750326
N
benth
am
iana
N m
uta
bilis
CP
G3 w
hite
N
muta
bilis
CP
G12456 w
hite
N
wig
andio
ides
N a
cum
inata
N fo
rste
riN
cle
vela
ndii
N n
octiflo
raN
x o
btu
sia
ta lin
e 2
N
x o
btu
sia
ta lin
e 1
N
mie
rsii
N p
etu
nio
ides
N x
obtu
sia
ta lin
e 5
N
sylv
estris
6898
N
oto
phora
white
N
are
nts
iiT
H3
2N
tabacum
lsquoChulu
manirsquo
N o
btu
sifo
lia v
ar p
alm
eri
N o
btu
sifo
lia v
ar o
btu
sifo
lia lsquoB
ald
win
rsquoN
linearis
TW
77
N n
udic
aulis
N o
btu
sifo
lia v
ar o
btu
sifo
lia T
W143
N lin
earis
964750099
N
pauciflo
raN
tom
ento
sifo
rmis
N o
tophora
pin
k
N g
lutin
osa
synth
etic
N ta
bacum
TH
37
N
tabacum
51789
synth
etic
N ta
bacum
QM
N
tabacum
095-5
5
N s
etc
hellii
N m
uta
bilis
CP
G3 p
ink
N m
uta
bilis
CP
G12456 p
ink
10
8
6
4
2
0
Dis
tan
ce
Spectral categories
MagentaRedPinkUV-white
WhiteYellowDark greenGreen
Dark greenSaturated greenGreenUV-whiteUV-green
PinkLight pinkWhiteUV-pinkRedSaturated UV-pink
SpecBeeHum
SpectralDark greenSaturated greenSaturated UV-blueUV-blueSaturated UV-greenHigh UV
UV-greenGreenLight greenBlueBlue-green
Bee Hummingbird
FIG 3 Dendrograms based on distance cluster analyses for spectral reflectance Coloured circles on the dendrogram represent distinct colour categories as determinedby the chosen threshold (dashed line) Dendrograms were similarly constructed for bee and hummingbird colour (see Supplementary Data Fig S3) The lines of col-oured circles at the tips of the dendrograms signify the category each taxon is assigned to in spectral bee and hummingbird colour as labelled (Spec Bee and Hum)for comparison between spectral categories and those of different visual systems Diploids polyploids and homoploids are denoted by black blue and orange text
respectively
1124 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
0
0middot2
0middot4
0middot6
0middot8
1middot0
Reflecta
nce
Wavelength (nm)
0
0middot2
0middot4
0middot6
0middot8
1middot0
300 350 400 450 500 550 600 650 700 300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
N benavidesii
N glauca 51751 g
N langsdorffii
N paniculata
N raimondii
N rustica var asiatica
N rustica var pavonii
synthetic UxP
synthetic F1
syn N rustica S0
syn N rustica S1
N undulata
N glutinosa
N otophora p
N tabacum 51789
syn N tabacum QM
syn N tabacum TH37
N tomentosiformis
A B
G 50
R 50
B 50
UV 50
Hummingbird colour space
White
0middot40middot4
ndash0middot40middot4
0middot4
0middot4
y
z
UV-white
Pink
Synthetic N tabacum QM
UV-pink
N tabacum 095-55
N setchellii
Green
N langsdorffii
Saturated green
UV-green
C DBee colour hexagon
Blue-green
Light green
Green
Dark green
Sat green
UV-green
Sat UV-green
B
B-G
G
UV-G
UV
UV-B
1 acum2 aren3 atten4 bena5 benth6 clev7 forst8 glau259 glau51y10 glau51g11 glut12 goss13 knight14 langs15 lin964716 linTW7717 mega18 mier19 mut1w20 mut1p21 mut3w22mut3p23 neso24 noct25 nudi26 xobtus127 xobtus228 xobtus529 obtusB30 obtusTW31 obtuspalm
32 occhesp33 otoph w34 otoph p35 pani36 pauc37 petun38 plumba39 quad904740 quadTW1841 raim42 repa43 rustasi44 rustpav45 syn UxP46 syn F147 syn rust S048 syn rust S149 setch50 stock51 suav52 sylv689853 sylvA04754 tab0955555 tab5178956 tabchulu57 syntabQM58 syntabTH3759 tomtform60 undu61 wigan62TH32
High UV
UV-blue
Sat UV-blue
Blue
FIG 4 (A B) Nicotiana reflectance spectra from 300 to 700nm which roughly correspond to colours perceived by human observers as pink (A) and green (B) SeeSupplementary Data Fig S2 for other spectral colour categories Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploidhybrid taxa Abbreviations p pink syn synthetic g green (C) Colour hexagon displaying the distribution of Nicotiana colour loci in bee colour space The hexa-gon has been scaled so that vertices represent 40 excitation of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-GUVndashgreen Bee colour categories are delineated by coloured ovals sat saturated Nicotiana species abbreviations are as follows acum acuminata aren arentsiiatten attenuata benavid benavidesii benth benthamiana clev clevelandii forst forsteri glau25 glauca 51725 glau51y glauca 51751 yellow glau51g glauca51751 green glut glutinosa goss gossei knight knightiana langs langsdorffii lin9647 linearis 964750099 linTW77 linearis TW77 mega megalosiphon miermiersii mut1w mutabilis CPG12456 white mut1p mutabilis CPG12456 pink mut3w mutabilis CPG3 white mut3p mutabilis CPG3 pink neso nesophila noctnoctiflora nudi nudicaulis obtus1 obtusiata line 1 obtus2 obtusiata line 2 obtus5 obtusiata line 5 obtusB obtusifolia var obtusifolia lsquoBaldwinrsquoobtusTW obtusifolia var obtusifolia TW143 obtuspalm obtusifolia var palmeri occhesp occidentalis subsp hesperis otoph w otophora white otoph p otophorapink pani paniculata pauc pauciflora petun petunioides plumba plumbaginifolia quad9047 quadrivalvis 904750042 quadTW18 quadrivalvis TW18 raimraimondii repa repanda rustasi rustica var asiatica rustpav rustica var pavonii syn UP synthetic UP syn F1 synthetic PUE1 F1 synrust S0 synthetic rus-tica PUE1-R10 S0 synrust S1 synthetic rustica PUE1-R1 S1 setch setchellii stock stocktonii suav suaveolens sylv6898 sylvestris 6898 sylvA047 sylvestrisA04750326 tab09555 tabacum 095-55 tab51789 tabacum 51789 tabchulu tabacum lsquoChulumanirsquo syntabQM synthetic tabacum QM syntabTH37 synthetictabacum TH37 tomtform tomentosiformis undu undulata wigan wigandioides TH32 TH32 synthetic N sylvestrisN otophora polyploid (D) Distribution ofNicotiana spectral loci in hummingbird colour space Vertices of the hummingbird colour space represent 50 excitation of the photoreceptors single photoreceptortype vertices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows repre-sent the vectors of these photoreceptors from the origin of the hummingbird colour space Nicotiana loci are coloured according to hummingbird colour categories(Supplementary Data Fig S3B) but are labelled with the accession name if the category includes only one taxon See Supplementary Data Video for an animation
of this graph
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1125
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
is the exception because it falls into an unexpected humming-bird colour category given the progenitor species (Fig 6)Section Repandae polyploids seem to have evolved to be eitherlike their maternal (Nicotiana nesophila N repanda andN stocktonii) or paternal (N nudicaulis) progenitor (Fig 6)The maternal progenitor N sylvestris is no longer sympatricwith any of the section Repandae polyploids therefore Nnesophila N repanda and N stocktonii can occupy the samepollination niche as their maternal progenitor without competi-tion Similarly section Suaveolentes is native to Australasiaexcept for one species in Namibia Africa and is not sympatricwith its progenitor sections in South America (Goodspeed1954) these polyploids and their diploid progenitors displaysimilar floral colours except N pauciflora which evolvedspectrally UVndashwhite flowers after the formation of sectionSuaveolentes (Fig 6) It is possible that there is less competitionfor pollinators and therefore reduced selective pressure to-wards floral colour diversification when polyploid species arenot sympatric with their diploid progenitors as is seen in
Iochrominae (Solanaceae) which have a broader range of floralcolours when species are sympatric (Muchhala et al 2014)However floral colour evolution can also be driven by geneticdrift or selection on pleiotropic effects of floral genes (Chittkaet al 2001 Rausher 2008) Furthermore anthocyanins andflavonoids are important for UV protection can defend againstfungi act as signal molecules and play a role in male fertility insome species in addition to their roles in signalling to pollina-tors (Shirley 1996)Over half of polyploids have an unexpected inheritance pat-
tern for chlorophyll if it is assumed that the presence of chloro-phyll is a dominant character Those polyploids that deviatefrom expectation span an age range of synthetic to 10myoand always lack chlorophyll in their petals which is in linewith the direction of the shifts observed in the divergence ofdiploid species (Supplementary Data Fig S6) In carnation thedifference in chlorophyll concentration between white andgreen flowers is likely caused by downregulation of chlorophyllbiosynthesis genes in white flowers genes involved in
TABLE 1 Polyploid and homoploid hybrid observed and expected floral colours
Species Spectral Bee Hummingbird Chlorophyll
Observed Expected Observed Expected Observed Expected Observed Expected
N tabacum 095-55 R W P UV-B B-G sUV-P W P N CN tabacum 51789 P W P B-G B-G P W P N CN tabacum lsquoChulumanirsquo W W P B-G B-G P W P N Csyn N tabacum QM P W P B B-G LP W P N Csyn N tabacum TH37 P W P B-G B-G P W P N CTH32 P W P B-G B-G P W P N CN rustica var asiatica G G LG G B-G G G W C CN rustica var pavonii G G G G B-G G G W C Csyn UP (homoploid) G G UV-G G B-G UV-G G W C Csyn F1 (homoploid) G G LG G B-G G G W C Csyn N rustica S0 G G LG G B-G G G W C Csyn N rustica S1 G G LG G B-G G G W C CN arentsii W G W B-G B-G P W C CN clevelandii W UV-W W B-G UV B-G P UV-W W N CN quadrivalvis TW18 W UV-W W B-G UV B-G W UV-W W C CN quadrivalvis 9047 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 1 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 2 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 5 W UV-W W B-G UV B-G W UV-W W C CN repanda W W UV-W B-G B-G UV W W UV-W N CN nesophila W W UV-W B-G B-G UV W W UV-W N CN stocktonii W W UV-W B-G B-G UV W W UV-W C CN nudicaulis UV-W W UV-W UV B-G UV UV-W W UV-W C CN benthamiana W W ndash ndash ndash ndash N CN forsteri W W ndash ndash ndash ndash C CN gossei W W ndash ndash ndash ndash N CN megalosiphon W W ndash ndash ndash ndash N CN occidentalis W W ndash ndash ndash ndash N CN suaveolens W W ndash ndash ndash ndash N C
N glauca 51725 Y W ndash ndash ndash ndash C CN glauca 51751 YG W ndash ndash ndash ndash C CN linearis TW77 UV-W W ndash ndash ndash ndash C CN linearis 9647 UV-W W ndash ndash ndash ndash C CN glutinosa P P W G ndash ndash ndash ndash C C
The top block consists of polyploid accessions and the bottom block includes homoploid hybridsItalic denotes an unexpected phenotype given the colour categories of the progenitorsProgenitor bee and hummingbird colour categories are unknown for section Suaveolentes and natural homoploid hybrids (see text)R red W white P pink G green UV-W UVndashwhite Y yellow UV-B UVndashblue B-G bluendashgreen B blue LG light green UV high UV UV-G
UVndashgreen sUV-P saturated UVndashpink LP light pink N no chlorophyll C chlorophyll syn synthetic 9047 represents 904750042 9647 represents964750099
1126 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
MATERIALS AND METHODS
Petal cell area measurements
To assess whether an increase in ploidy results in larger petalcells cell area was measured from a subset of polyploids andtheir progenitors For Nicotiana sylvestris A04750326Nicotiana rustica var asiatica Nicotiana rustica var pavoniiNicotiana paniculata Nicotiana undulata and Nicotiana nudi-caulis mature flowers were taken from plants and the adaxialpetal surface was imprinted in Elite HD vinylpolysiloxane im-pression material (dental wax supplied by ZhermackHarrogate UK) The wax was left to set and then used as amould for making epoxy petal casts Devcon high-strength ep-oxy was mixed according to the manufacturerrsquos instructionspoured into the mould and allowed to set for 12 h The epoxyrelief was removed and coated with gold using a QuorumK756X sputter coater The samples were then imaged using aFEI Philips XL30 FEGSEM scanning electron microscope ForNicotiana obtusifolia var obtusifolia TW143 Nicotianarepanda and Nicotiana stocktonii only fixed material wasavailable whole mature flowers were fixed in formalinndashaceticacidndashalcohol (FAA) (60 ethanol 6 formaldehyde 5 acetic acid) for 72h before being transferred to a 70 ethanol(EtOH) wash for 24 h The samples were then dehydratedthrough an ethanol series of 2 h each in 70 80 and 90 andtwo washes in 100 EtOH The samples were dissected andthen dried in an Autosamdri 815B critical point dryer Thesesamples were sputter-coated and imaged as described aboveFor all samples images were taken mid-petal from an angleperpendicular to the surface to minimize parallax error Cellsize measurements were carried out in ImageJ (httpimagejnihgovij) The circumference of the cell base was drawn free-hand and area was calculated for 100ndash150 cells until the cu-mulative mean stabilized One-way ANOVA and Tukeyrsquoshonest significance tests were performed in RStudio version098490 (httpwwwrstudioorg) to compare cell area of poly-ploids with those of their progenitors repeating the tests foreach polyploid section
Spectral reflectance measurements
Spectral reflectance measurements were recorded for 60Nicotiana accessions (41 taxa Supplementary Data Table S1)three flowers from different plants where possible were usedfor each accession Reflectance spectra from three Nicotianaotophora accessions were pooled because the spectra weresimilarSpectral reflectance of flowers at anthesis was measured
from 300 to 700 nm using an Avantes AvaSpec-2048 spectro-photometer with an Avantes AvaLight-DHS light source andcalibrated with a barium sulphate white standard from lab-sphereVR Nicotiana mutabilis was also measured later as flow-ers change from white to pink when mature pink flowers areless likely to have a nectar reward but add to the attraction ofthe overall floral display and therefore are still relevant to pol-linators (R Kaczorowski University of Haifa Israel perscomm) Reflectance spectra express the proportion of lightreflected by the flower at any given wavelength Spectra werevisualized and exported in increments of 1 nm using the
program AvaSoft version 703 Full (Avantes BV Eerbeek TheNetherlands) and imported into ExcelSpectra for each accession or colour morph were averaged
and then smoothed three times using a rolling average over9 nm Spectra for all accessions were submitted to the FloralReflectance Database (FReD wwwreflectancecouk Arnoldet al 2010)Some spectra had a spike at 656 nm which corresponded
to a narrow peak in the light source spectrum suggesting thatthe spectra were saturated at 656 nm however smoothingserved to neutralize this spike Several spectra (Nicotiana are-ntsii N mutabilis Nicotiana suaveolens and Nicotiana wigan-dioides) included an anomalous reflectance minimum from 475to 500 nm which could not be explained by the light sourcespectrum Remeasured spectra of N arentsii N suaveolens andN wigandioides lacked this minimum but further material ofN mutabilis was unavailable so these spectra were includeddespite the anomalies
Calculation of colour loci in the bee colour hexagon
A reflectance spectrum can be represented as a single pointin the bee colour hexagon space (a graphical representation of abeersquos colour visual experience) based on the relative excitationof UV- blue- and green-sensitive photoreceptor types (Chittka1992) Vertices of this hexagon represent theoretical stateswhere one or two photoreceptor types are at maximal excitationwhereas at least one receptor type is at zero excitation (eg thetop vertex of the hexagon corresponds to maximal blue receptorexcitation and zero signal from UV and green receptorswhereas the top right vertex corresponds to maximal signal inboth blue and green receptors but no signal in the UV receptorand so forth see Supplementary Data Fig S1) The centre ororigin of the hexagon is achromatic Hue corresponds to angu-lar position around the origin whereas spectral purity or satura-tion increases with distance from the originBee colour hexagon coordinates were calculated for all
Nicotiana spectra Illumination was assumed to be sunlight(D65 Wyszecki and Stiles 1982) the background was repre-sented by an average leaf spectrum (Gumbert et al 1999)Honeybee photoreceptor spectral sensitivity functions wereused to determine bee colour hexagon coordinates these aresimilar to bumblebee and hawkmoth photoreceptor sensitivityfunctions so the bee colour hexagon can be used to approxi-mate the colour vision of these insects as well (Menzel et al1986 Peitsch et al 1992 Briscoe and Chittka 2001 Kelberet al 2003 and references therein Skorupski et al 2007) Theequations used to determine colour hexagon coordinates are asfollows where EG EB and EUV represent the excitation of thegreen blue and UV bee photoreceptors respectively elicitedby a spectrum (Chittka 1992)
x frac14ffiffiffi
3p
=2ethEG EUVTHORNy frac14 EB 05ethEUV thorn EGTHORN
Because the colour loci of Nicotiana flowers were mostlyclose to the centre of the colour space all colour hexagon dis-plays presented are scaled so that only the central 40 is
1120 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
shown the outline is therefore drawn as a dashed line This re-sults in a clearer spread of the colour loci to facilitate visual in-spection For a diagram explaining the colour hexagon seeSupplementary Data Fig S1
Calculation of colour loci in hummingbird colour space
For tetrachromatic hummingbirds we chose to model flowercolours in a 3-D colour opponent space because n ndash 1 colouropponent dimensions are necessary to code the informationfrom n colour receptors (Chittka 1996) The hummingbirdcolour space can be displayed as a rhombic dodecahedron with14 vertices (Restrepo 2013) Like the bee colour hexagonvertices of the space represent states where one two or threephotoreceptor types are at maximal excitation and at least onereceptor type is at zero excitation
Hummingbird colour space coordinates were calculated forall Nicotiana spectra Illumination was again assumed to besunlight (D65 Wyszecki and Stiles 1982) and the backgroundan average leaf spectrum (Gumbert et al 1999) as was usedfor the bee colour hexagon Photoreceptor spectral sensitivityfunctions from the green-backed firecrown hummingbird(Sephanoides sephanoides Herrera et al 2008) were used todetermine hummingbird colour space coordinates usingthe following equations (Restrepo 2013) where ER EGEB and EUV represent excitation of red green blue andUV hummingbird photoreceptors respectively elicited by aspectrum
x frac14ffiffiffiffiffiffiffiffi
3=4p
EB ffiffiffiffiffiffiffiffiffiffi
1=12p
ethEUV thorn EG thorn ERTHORN
y frac14ffiffiffiffiffiffiffiffi
2=3p
EG ffiffiffiffiffiffiffiffi
1=6p
ethEUV thorn ERTHORN
z frac14ffiffiffiffiffiffiffiffi
1=2p
ethEUV ERTHORN
RStudio was used to make 3D plots of the hummingbird col-our space and ImageJ version 148 (httpimagejnihgovij)was used to create an animation of the Nicotiana flower loci inthe hummingbird colour space Again Nicotiana flower colourloci are close to the origin in the hummingbird colour space sothe graphs presented display only the central portion (either 25or 50 ) of the colour space for clarity To further facilitateinterpretation of these graphs vertices representing individualexcitation of the red green blue and UV photoreceptor typesas well as their excitation vectors from the origin are shownin red green blue and black respectively Other vertices(representing excitation of two or three photoreceptor types) areshown in grey
Cluster analyses
Cluster analyses were used to group spectra based on spectralshape (corresponding to pigmentation) and their position inboth bee and hummingbird colour spaces For spectral colourcategories spectra were normalized to the same integral underthe curve in order to compare combinations of pigments notthe concentration of pigments A distance matrix was calculatedfrom the normalized spectral data in R version 302 (http
wwwR-projectorg) using the dist() function For the bee andhummingbird colour categories the input data were the (x y) or(x y z) coordinates of the spectra in the bee and hummingbirdcolour spaces respectivelyData were first imported into R The function hclust() was
used to perform agglomerative hierarchical clustering based onthe average pairwise distances between groups With this algo-rithm the observed points which are initially all deemed to bedistinct are iteratively assigned to groups until eventually allpoints belong to the same group At each step the average dis-tance between all groups (ie the mean distance from all pointsin group A to all points in group B if either one of these is asingle point then no averaging is needed) is calculated and thetwo groups with the smallest average distance are merged Theorder in which groups are merged can be used to construct adendrogram showing the spatial relationship between all datapoints We can also look at the distribution of merge distancesat each step in the algorithm and use this distribution to esti-mate how many groups are present in the data Points at whichthere is a steep increase in the average between-group distance(lsquoelbowrsquo points) highlight the spatial scale(s) at which there isclustering present in the data By using one of these elbowpoints as a cutoff in the algorithm we can arrive at a distancegrouping that captures the spatial clustering It should be notedthat the determination of where to draw the threshold in a clus-ter analysis is arbitrary but the use of one of these elbow pointsdoes yield meaningful clusters The determination of the spe-cific point from the elbow region to be used to define clusterswas further informed by visual inspection of reflectance spec-tra as well as distributions of colour loci in the perceptual col-our spaces It should be noted that the dendrograms relate tosimilarities in spectral reflectance as well as colour relation-ships perceived by bees and hummingbirds they do not showphylogenetic relationships
Ancestral state reconstruction
To examine the evolution of colour within a phylogeneticcontext ancestral state reconstructions were performed on treesinferred from plastid sequence data Only species for which flo-ral character data are available were included in these analysesBecause polyploid and homoploid hybrids originate via reticu-late evolutionary processes and therefore lack a history oftree-like evolution ancestral characters were reconstructed us-ing only non-hybrid diploid species The states observed inhybrid species were then compared with the ancestral state re-constructions Since sections Repandae and Suaveolentes havediversified to form several species following polyploidizationcharacters were reconstructed for these sections separately toexamine colour evolution subsequent to their origin Fornon-hybrid diploid species individual gene trees yieldsome conflicting topologies nevertheless key nodes for thepurposes of interpreting character evolution in hybrids arerecovered in multiple gene trees and are supported bysupernetwork analyses (Kelly et al 2010) Therefore plastiddata from previously published studies are suitable for theseanalysesPreviously published sequences (Clarkson et al 2004) from
four plastid regions (matK ndhF trnL-F and trnS-G) were
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1121
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
aligned separately using PRANKthornF (Loytynoja and Goldman2008) and then concatenated to create a combined plastid data-set before further optimization by eye in Mesquite version 274(Maddison and Maddison 2008) For Nicotiana attenuata weused GenBank accessions AB040009 and AY098697 for thematK and trnL-F regions respectively (due to likely misidenti-fication of N attenuata material used in Clarkson et al 2004see Clarkson et al 2010) the other two regions were scored asmissing data Phylogenetic reconstruction by Bayesian infer-ence was performed as described in Kelly et al (2013) with theexception that BayesTrees v13 (wwwevolutionreadingacukBayesTreeshtml) was used to construct 95 majority ruleconsensus trees For sections Repandae and Suaveolentes se-quences representing their putative maternal progenitors wereincluded during Bayesian inference to allow rooting of trees butwere pruned from the trees prior to ancestral statereconstructionAncestral states for spectral reflectance colour categories and
presenceabsence of chlorophyll in petals (data inSupplementary Data Table S2) were reconstructed using theparsimony reconstruction method in Mesquite version 274 un-der the unordered states assumption To account for topologicaluncertainty character states were reconstructed over all 36 000post-burn-in trees using the Trace Character Over Trees optionand summarized on the 95 majority rule consensus tree fromthe Bayesian analysis Ancestral states were not calculated forbee or hummingbird colour categories because these are per-ceptual systems and the same colour category can result fromdifferent combinations of pigments (eg both human pink andhuman white flowers which are positive and negative respec-tively for anthocyanin pigmentation are both classified as beebluendashgreen) thus a single colour category does not necessarilyhave a shared evolutionary history
Estimating expected polyploid and homoploid hybrid floralcolour
Polyploid and homoploid hybrid floral colours for each ac-cession were compared with those of their diploid progenitorsfor the spectral bee and hummingbird colour categories definedby cluster analyses Floral colour was classified as lsquoexpectedrsquo ifit fell in the colour category of at least one progenitor or lsquounex-pectedrsquo if it was different from both progenitors Polyploid andhomoploid hybrids were also compared with their diploid pro-genitors for the presence or absence of chlorophyll in corollatissue Chlorophyll absorbs at 675 nm in vivo (Haardt andMaske 1987) therefore the presence of chlorophyll can be in-ferred from reflectance spectra if there is a reflectance mini-mum at 675 nm The presence of chlorophyll in petal tissueappears to be dominant (Brieger 1935 Ohmiya et al 2014)Thus hybrids were classified as expected if they showed chlo-rophyll pigmentation and as unexpected if they did not becauseat least one progenitor possessed chlorophyll in all diploid pro-genitor combinations For natural homoploid hybrids and poly-ploid section Suaveolentes where progenitors can only bedefined to Nicotiana section level comparisons were madewith reconstructed ancestral characters thus only spectral col-our categories and the presenceabsence of chlorophyll wereexamined
Phylogenetic signal in floral traits
In order to statistically test for phylogenetic signal in the phe-notypic trait data (spectral reflectance bee and hummingbirdcolour perception) we used Mantel tests to examine the corre-lation between phylogenetic distance and each of the respectivecontinuous multidimensional traits (eg Cubo et al 2005Muchhala et al 2014) Analyses were restricted to diploid spe-cies excluding homoploid and polyploid hybrids Trees wereedited in Newick format to include additional tips with zerobranch lengths for taxa that are multiple in the trait datasetseither due to colour polymorphism (N otophora) or multipleaccessions (N sylvestris and N obtusifolia var obtusifolia)Statistical analyses were performed in R version 310
Phenotypic distance matrices were first calculated for the threetrait datasets using Euclidean distance and phylogenetic dis-tance matrices were calculated (1) as genetic distance from theplastid alignment and (2) for each of 36 000 post-burn-inBayesian trees using copheneticphylo() part of the APE pack-age version 31-2 (Paradis et al 2004) The second Bayesianset of tests was performed in order to account for evolutionaryprocesses such as saturation and to estimate how phylogeneticuncertainty affects the correlation Mantel tests were performedusing Pearsonrsquos product-moment correlation coefficient with10 000 permutations of each distance matrix to test for signifi-cance the mean P value and its standard deviation were calcu-lated for each set of 36 000 Mantel tests from the Bayesiantrees along with the percentage of trees that gave significantcorrelations The function mantel() from the vegan packagewas used (Oksanen et al 2013)
RESULTS
Petal cell area
Petal cell area was measured to determine whether an increasein ploidy results in larger floral cells Polyploid petal cell areawas significantly larger than in both progenitors in N rustica(ANOVA Ffrac14 371 dffrac14 3 Plt 2 10ndash16) accessions but wasintermediate between progenitors in section Repandae poly-ploids (ANOVA Ffrac14 2492 dffrac14 4 Plt 2 10ndash16 Fig 2)Accessions that were significantly different in cell area (withinpolyploid sections and their progenitors) are represented bydifferent letters above the bars in Fig 2 results from Tukeyrsquoshonest significance tests can be found in Supplementary DataTable S3
Cluster analyses
Nicotiana reflectance spectra were grouped into categoriesbased on spectral shape and position in the bee and humming-bird colour spaces using cluster analyses Bees and humming-birds have different photoreceptor sensitivities and we expectour cluster analyses to reflect these differences in sensoryequipment The analysis based on spectral shape yielded eightcolour categories which roughly corresponded to flowers per-ceived by human observers as magenta red pink UVndashwhitewhite yellow green and dark green (Fig 3) Nicotiana spectraare displayed by spectral colour category in Fig 4A BSupplementary Data Fig S2 The bee colour hexagon clustering
1122 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
resulted in 11 colour categories which fell into the followingareas of bee colour space saturated green UVndashblue high UVUVndashgreen green light green bluendashgreen dark green saturatedUVndashblue saturated UVndashgreen and blue (the last four categorieswere each represented by only a single accessionSupplementary Data Fig S3A) These groups are shown in thebee colour hexagon (Fig 4C) The hummingbird colour spacecluster analysis also produced 11 colour categories saturatedgreen green UVndashwhite UVndashgreen pink white UVndashpinkdark green light pink red and saturated UVndashpink (again thelast four categories include only a single accessionSupplementary Data Fig S3B) These groups are shown in thehummingbird colour space (Fig 4D) and the same graph isprovided as an animation to better display the 3D nature of thecolour space (Supplementary Data Video)
Evolution of spectral reflectance in polyploids and homoploidhybrids
To assess the evolution of polyploid floral colour polyploidand homoploid hybrid accessions were compared with those oftheir progenitors in spectral bee and hummingbird colour cate-gories as well as in the presenceabsence of chlorophyll Thediploid progenitors and approximate age of polyploids andhomoploid hybrids are found in Fig 1 and Supplementary DataTable S4 and the observed and expected floral colours of
polyploid and homoploid hybrids are found in Table 1 Mostpolyploid and homoploid hybrids were similar to at least oneprogenitor in spectral bee and hummingbird colour categoriesbut some fell into unexpected colour categories (Table 1Fig 5 Supplementary Data Fig S4 Supplementary Data FigS5) Over half of the polyploids unexpectedly lacked chloro-phyll (Table 1)
Evolution of colour characters in a phylogenetic context
Reconstructed character states are shown for spectral reflec-tance colour categories (Fig 6) and the presenceabsence ofchlorophyll in petals (Supplementary Data Fig S6) Bee andhummingbird colour categories are also shown for extant spe-cies on the plastid tree (Fig 6) Although the deepest nodeswere largely equivocal evolution of spectral reflectance colourin Nicotiana seemed to be dynamic (Fig 6) Green flowerslikely have three independent origins (1) in sectionsPaniculatae and Undulatae (2) in N langsdorffii and (3) inthe homoploid hybrid N glauca UVndashwhite flowers also seemto have arisen three times independently (1) in sectionTrigonophyllae (2) in N pauciflora and (3) in the homoploidhybrid N linearis Most polyploid and homoploid hybrid spe-cies exhibit a floral colour present in at least one of their pro-genitors However N tabacum 095-55 is red and N glauca isyellow and green unlike their progenitors UVndashwhite flowersseem to have evolved de novo in N linearis UVndashwhite flowersare also found in one of its progenitor sections (in N pauci-flora) but ancestral reconstructions indicate that the floral col-our was most likely white at the ancestral nodes within thesection (Fig 6) This suggests that the evolution of UVndashwhiteflowers in N pauciflora has occurred subsequent to the forma-tion of N linearis and that the two events are likely indepen-dent It is unclear whether UVndashwhite flowers also evolved denovo in N nudicaulis because the ancestral node of sectionRepandae is equivocal The presence of chlorophyll as inferredby light absorption at 675 nm (Haardt and Maske 1987) inNicotiana flowers is ancestral and has been lost three times inN sylvestris N noctiflora and the most recent common ances-tor of N acuminata and N pauciflora (Supplementary DataFig S6)Results from Mantel tests of phylogenetic signal for
Nicotiana floral traits for both genetic distance and the 36 000post-burn-in Bayesian trees are shown in Table 2 All floraltraits were significantly correlated with phylogenetic relation-ships for the Bayesian trees at a significance level of Plt 005Only spectral reflectance was significant for the genetic dis-tance tests whereas bee and hummingbird colour perceptionwere just above the Plt 005 threshold For the Bayesian trees901 662 and 932 of trees were significantly correlatedwith the spectral reflectance bee and hummingbird colour per-ception datasets respectively These results suggest that thesefloral traits are weakly constrained by phylogeny
DISCUSSION
Nicotiana is remarkable in its range of spectral reflectanceflower colours (white UVndashwhite pink magenta red yellowgreen and dark green Fig 3) and in the variety of pollinators
0
200
400
600
800
1000
1200C
ell
are
a (
microm
2)
N rustica
N paniculata
N undulata
N sylvestris
N obtusifolia
N nudicaulis
N repanda
N stocktonii
N rustica var asiatica
N rustica var pavonii
Repandae
a
a
b
b
cc
c c
d
Maternal progenitor
Paternal progenitor
Polyploids
Polyploids
Polyploids
FIG 2 Petal cell area from polyploids and their progenitors Within each poly-ploid group bars with different letters represent significantly different mean cell
areas
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1123
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
that visit the flowers (moth bird bee bat Knapp 2010) Theperception of these spectral colours also changes with visualsystem (bee or hummingbird) Here we describe a complex dy-namic in the evolution of floral colour in Nicotiana Spectral re-flectance and bee and hummingbird colour perception arecorrelated with phylogeny but multiple independent origins ofvarious combinations of pigmentation suggest that the evolutionof floral colour is not entirely phylogenetically constrained
Petal cell size evolution in polyploids
Cell size is expected to increase following polyploidizationdue to the increase in genome size (Beaulieu et al 2008) Thesignificantly larger petal cells of N rustica (lt02 million yearsold myo) and the intermediate cell size of section Repandae(45myo) suggest that polyploids may revert to a diploid-like cell size over time similar to the genome downsizing ob-served in polyploids (Leitch and Bennett 2004) Howeverpetal cell size differences within section Repandae do not seemto be linked to genome size N nudicaulis and N repanda sharesimilar cell sizes but have substantially different genome sizes(Leitch et al 2008) The concentration of pigment in floralcells is controlled by both the amount of pigment present andthe cell size Nicotiana rustica has significantly larger petalcells than its progenitors (close to the sum Fig 2) and displaysan intermediate brightness (the area under the reflectance curvea proxy for pigment concentration) between its progenitors
(Supplementary Data Fig S4D) This is expected if the poly-ploid inherits the sum of both cell size and amount of pigmentpresent from its progenitors
Polyploid divergence in floral colour
Many younger polyploids (lt02myo) display unexpectedfloral colours considering those of their diploid progenitorsNone of the natural and synthetic N tabacum accessions pos-sess chlorophyll which is unexpected given its presence in atleast one progenitor species Nicotiana tabacum 095-55 alsohas unexpected spectral bee and hummingbird colour giventhe colour categories of the progenitor species Similarly syn-thetic N tabacum QM has unexpected colours in bee and hum-mingbird perception Because this accession is synthetic theparents are known and thus its unexpected phenotype can beclassified as transgressive or outside the range of its progeni-tors due solely to polyploidy and hybridization Most N rusticaaccessions have unexpected bee colour (four of these aresynthetic and are therefore also transgressive) and N arentsiihas unexpected hummingbird colour (Table 1 Fig 6) Despitethe divergence of floral spectra associated with polyploidybehavioural studies are needed to determine whether the beeand hummingbird colour categories delineated here actuallyelicit different responses in pollinatorsMost older polyploids (1ndash10myo) are similar in floral col-
our category to at least one of their progenitors N clevelandii
synth
etic
N ru
stic
a P
UE
1-R
1 S
1
N ru
stic
a v
ar a
sia
tica
synth
etic
N ru
stic
a P
UE
1-R
10 S
0
synth
etic
PU
E1 F
1synth
etic
UxP
N b
enavid
esii
N ru
stic
a v
ar p
avonii
N p
anic
ula
taN
langsdorffii
N g
lauca 5
1751 g
reen
N u
ndula
taN
raim
ondii
N k
nig
htia
na
N g
lauca 5
1751 y
ello
w
N g
lauca 5
1725
N p
lum
bagin
ifolia
N g
ossei
N q
uadriv
alv
is T
W18
N o
ccid
enta
lis s
ubsp h
esperis
N
megalo
sip
hon
N s
uaveole
ns
N n
esophila
N s
tockto
nii
N q
uadriv
alv
is 9
04750042
N a
ttenuata
N re
panda
N s
ylv
estris
A04750326
N
benth
am
iana
N m
uta
bilis
CP
G3 w
hite
N
muta
bilis
CP
G12456 w
hite
N
wig
andio
ides
N a
cum
inata
N fo
rste
riN
cle
vela
ndii
N n
octiflo
raN
x o
btu
sia
ta lin
e 2
N
x o
btu
sia
ta lin
e 1
N
mie
rsii
N p
etu
nio
ides
N x
obtu
sia
ta lin
e 5
N
sylv
estris
6898
N
oto
phora
white
N
are
nts
iiT
H3
2N
tabacum
lsquoChulu
manirsquo
N o
btu
sifo
lia v
ar p
alm
eri
N o
btu
sifo
lia v
ar o
btu
sifo
lia lsquoB
ald
win
rsquoN
linearis
TW
77
N n
udic
aulis
N o
btu
sifo
lia v
ar o
btu
sifo
lia T
W143
N lin
earis
964750099
N
pauciflo
raN
tom
ento
sifo
rmis
N o
tophora
pin
k
N g
lutin
osa
synth
etic
N ta
bacum
TH
37
N
tabacum
51789
synth
etic
N ta
bacum
QM
N
tabacum
095-5
5
N s
etc
hellii
N m
uta
bilis
CP
G3 p
ink
N m
uta
bilis
CP
G12456 p
ink
10
8
6
4
2
0
Dis
tan
ce
Spectral categories
MagentaRedPinkUV-white
WhiteYellowDark greenGreen
Dark greenSaturated greenGreenUV-whiteUV-green
PinkLight pinkWhiteUV-pinkRedSaturated UV-pink
SpecBeeHum
SpectralDark greenSaturated greenSaturated UV-blueUV-blueSaturated UV-greenHigh UV
UV-greenGreenLight greenBlueBlue-green
Bee Hummingbird
FIG 3 Dendrograms based on distance cluster analyses for spectral reflectance Coloured circles on the dendrogram represent distinct colour categories as determinedby the chosen threshold (dashed line) Dendrograms were similarly constructed for bee and hummingbird colour (see Supplementary Data Fig S3) The lines of col-oured circles at the tips of the dendrograms signify the category each taxon is assigned to in spectral bee and hummingbird colour as labelled (Spec Bee and Hum)for comparison between spectral categories and those of different visual systems Diploids polyploids and homoploids are denoted by black blue and orange text
respectively
1124 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
0
0middot2
0middot4
0middot6
0middot8
1middot0
Reflecta
nce
Wavelength (nm)
0
0middot2
0middot4
0middot6
0middot8
1middot0
300 350 400 450 500 550 600 650 700 300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
N benavidesii
N glauca 51751 g
N langsdorffii
N paniculata
N raimondii
N rustica var asiatica
N rustica var pavonii
synthetic UxP
synthetic F1
syn N rustica S0
syn N rustica S1
N undulata
N glutinosa
N otophora p
N tabacum 51789
syn N tabacum QM
syn N tabacum TH37
N tomentosiformis
A B
G 50
R 50
B 50
UV 50
Hummingbird colour space
White
0middot40middot4
ndash0middot40middot4
0middot4
0middot4
y
z
UV-white
Pink
Synthetic N tabacum QM
UV-pink
N tabacum 095-55
N setchellii
Green
N langsdorffii
Saturated green
UV-green
C DBee colour hexagon
Blue-green
Light green
Green
Dark green
Sat green
UV-green
Sat UV-green
B
B-G
G
UV-G
UV
UV-B
1 acum2 aren3 atten4 bena5 benth6 clev7 forst8 glau259 glau51y10 glau51g11 glut12 goss13 knight14 langs15 lin964716 linTW7717 mega18 mier19 mut1w20 mut1p21 mut3w22mut3p23 neso24 noct25 nudi26 xobtus127 xobtus228 xobtus529 obtusB30 obtusTW31 obtuspalm
32 occhesp33 otoph w34 otoph p35 pani36 pauc37 petun38 plumba39 quad904740 quadTW1841 raim42 repa43 rustasi44 rustpav45 syn UxP46 syn F147 syn rust S048 syn rust S149 setch50 stock51 suav52 sylv689853 sylvA04754 tab0955555 tab5178956 tabchulu57 syntabQM58 syntabTH3759 tomtform60 undu61 wigan62TH32
High UV
UV-blue
Sat UV-blue
Blue
FIG 4 (A B) Nicotiana reflectance spectra from 300 to 700nm which roughly correspond to colours perceived by human observers as pink (A) and green (B) SeeSupplementary Data Fig S2 for other spectral colour categories Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploidhybrid taxa Abbreviations p pink syn synthetic g green (C) Colour hexagon displaying the distribution of Nicotiana colour loci in bee colour space The hexa-gon has been scaled so that vertices represent 40 excitation of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-GUVndashgreen Bee colour categories are delineated by coloured ovals sat saturated Nicotiana species abbreviations are as follows acum acuminata aren arentsiiatten attenuata benavid benavidesii benth benthamiana clev clevelandii forst forsteri glau25 glauca 51725 glau51y glauca 51751 yellow glau51g glauca51751 green glut glutinosa goss gossei knight knightiana langs langsdorffii lin9647 linearis 964750099 linTW77 linearis TW77 mega megalosiphon miermiersii mut1w mutabilis CPG12456 white mut1p mutabilis CPG12456 pink mut3w mutabilis CPG3 white mut3p mutabilis CPG3 pink neso nesophila noctnoctiflora nudi nudicaulis obtus1 obtusiata line 1 obtus2 obtusiata line 2 obtus5 obtusiata line 5 obtusB obtusifolia var obtusifolia lsquoBaldwinrsquoobtusTW obtusifolia var obtusifolia TW143 obtuspalm obtusifolia var palmeri occhesp occidentalis subsp hesperis otoph w otophora white otoph p otophorapink pani paniculata pauc pauciflora petun petunioides plumba plumbaginifolia quad9047 quadrivalvis 904750042 quadTW18 quadrivalvis TW18 raimraimondii repa repanda rustasi rustica var asiatica rustpav rustica var pavonii syn UP synthetic UP syn F1 synthetic PUE1 F1 synrust S0 synthetic rus-tica PUE1-R10 S0 synrust S1 synthetic rustica PUE1-R1 S1 setch setchellii stock stocktonii suav suaveolens sylv6898 sylvestris 6898 sylvA047 sylvestrisA04750326 tab09555 tabacum 095-55 tab51789 tabacum 51789 tabchulu tabacum lsquoChulumanirsquo syntabQM synthetic tabacum QM syntabTH37 synthetictabacum TH37 tomtform tomentosiformis undu undulata wigan wigandioides TH32 TH32 synthetic N sylvestrisN otophora polyploid (D) Distribution ofNicotiana spectral loci in hummingbird colour space Vertices of the hummingbird colour space represent 50 excitation of the photoreceptors single photoreceptortype vertices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows repre-sent the vectors of these photoreceptors from the origin of the hummingbird colour space Nicotiana loci are coloured according to hummingbird colour categories(Supplementary Data Fig S3B) but are labelled with the accession name if the category includes only one taxon See Supplementary Data Video for an animation
of this graph
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1125
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
is the exception because it falls into an unexpected humming-bird colour category given the progenitor species (Fig 6)Section Repandae polyploids seem to have evolved to be eitherlike their maternal (Nicotiana nesophila N repanda andN stocktonii) or paternal (N nudicaulis) progenitor (Fig 6)The maternal progenitor N sylvestris is no longer sympatricwith any of the section Repandae polyploids therefore Nnesophila N repanda and N stocktonii can occupy the samepollination niche as their maternal progenitor without competi-tion Similarly section Suaveolentes is native to Australasiaexcept for one species in Namibia Africa and is not sympatricwith its progenitor sections in South America (Goodspeed1954) these polyploids and their diploid progenitors displaysimilar floral colours except N pauciflora which evolvedspectrally UVndashwhite flowers after the formation of sectionSuaveolentes (Fig 6) It is possible that there is less competitionfor pollinators and therefore reduced selective pressure to-wards floral colour diversification when polyploid species arenot sympatric with their diploid progenitors as is seen in
Iochrominae (Solanaceae) which have a broader range of floralcolours when species are sympatric (Muchhala et al 2014)However floral colour evolution can also be driven by geneticdrift or selection on pleiotropic effects of floral genes (Chittkaet al 2001 Rausher 2008) Furthermore anthocyanins andflavonoids are important for UV protection can defend againstfungi act as signal molecules and play a role in male fertility insome species in addition to their roles in signalling to pollina-tors (Shirley 1996)Over half of polyploids have an unexpected inheritance pat-
tern for chlorophyll if it is assumed that the presence of chloro-phyll is a dominant character Those polyploids that deviatefrom expectation span an age range of synthetic to 10myoand always lack chlorophyll in their petals which is in linewith the direction of the shifts observed in the divergence ofdiploid species (Supplementary Data Fig S6) In carnation thedifference in chlorophyll concentration between white andgreen flowers is likely caused by downregulation of chlorophyllbiosynthesis genes in white flowers genes involved in
TABLE 1 Polyploid and homoploid hybrid observed and expected floral colours
Species Spectral Bee Hummingbird Chlorophyll
Observed Expected Observed Expected Observed Expected Observed Expected
N tabacum 095-55 R W P UV-B B-G sUV-P W P N CN tabacum 51789 P W P B-G B-G P W P N CN tabacum lsquoChulumanirsquo W W P B-G B-G P W P N Csyn N tabacum QM P W P B B-G LP W P N Csyn N tabacum TH37 P W P B-G B-G P W P N CTH32 P W P B-G B-G P W P N CN rustica var asiatica G G LG G B-G G G W C CN rustica var pavonii G G G G B-G G G W C Csyn UP (homoploid) G G UV-G G B-G UV-G G W C Csyn F1 (homoploid) G G LG G B-G G G W C Csyn N rustica S0 G G LG G B-G G G W C Csyn N rustica S1 G G LG G B-G G G W C CN arentsii W G W B-G B-G P W C CN clevelandii W UV-W W B-G UV B-G P UV-W W N CN quadrivalvis TW18 W UV-W W B-G UV B-G W UV-W W C CN quadrivalvis 9047 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 1 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 2 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 5 W UV-W W B-G UV B-G W UV-W W C CN repanda W W UV-W B-G B-G UV W W UV-W N CN nesophila W W UV-W B-G B-G UV W W UV-W N CN stocktonii W W UV-W B-G B-G UV W W UV-W C CN nudicaulis UV-W W UV-W UV B-G UV UV-W W UV-W C CN benthamiana W W ndash ndash ndash ndash N CN forsteri W W ndash ndash ndash ndash C CN gossei W W ndash ndash ndash ndash N CN megalosiphon W W ndash ndash ndash ndash N CN occidentalis W W ndash ndash ndash ndash N CN suaveolens W W ndash ndash ndash ndash N C
N glauca 51725 Y W ndash ndash ndash ndash C CN glauca 51751 YG W ndash ndash ndash ndash C CN linearis TW77 UV-W W ndash ndash ndash ndash C CN linearis 9647 UV-W W ndash ndash ndash ndash C CN glutinosa P P W G ndash ndash ndash ndash C C
The top block consists of polyploid accessions and the bottom block includes homoploid hybridsItalic denotes an unexpected phenotype given the colour categories of the progenitorsProgenitor bee and hummingbird colour categories are unknown for section Suaveolentes and natural homoploid hybrids (see text)R red W white P pink G green UV-W UVndashwhite Y yellow UV-B UVndashblue B-G bluendashgreen B blue LG light green UV high UV UV-G
UVndashgreen sUV-P saturated UVndashpink LP light pink N no chlorophyll C chlorophyll syn synthetic 9047 represents 904750042 9647 represents964750099
1126 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
shown the outline is therefore drawn as a dashed line This re-sults in a clearer spread of the colour loci to facilitate visual in-spection For a diagram explaining the colour hexagon seeSupplementary Data Fig S1
Calculation of colour loci in hummingbird colour space
For tetrachromatic hummingbirds we chose to model flowercolours in a 3-D colour opponent space because n ndash 1 colouropponent dimensions are necessary to code the informationfrom n colour receptors (Chittka 1996) The hummingbirdcolour space can be displayed as a rhombic dodecahedron with14 vertices (Restrepo 2013) Like the bee colour hexagonvertices of the space represent states where one two or threephotoreceptor types are at maximal excitation and at least onereceptor type is at zero excitation
Hummingbird colour space coordinates were calculated forall Nicotiana spectra Illumination was again assumed to besunlight (D65 Wyszecki and Stiles 1982) and the backgroundan average leaf spectrum (Gumbert et al 1999) as was usedfor the bee colour hexagon Photoreceptor spectral sensitivityfunctions from the green-backed firecrown hummingbird(Sephanoides sephanoides Herrera et al 2008) were used todetermine hummingbird colour space coordinates usingthe following equations (Restrepo 2013) where ER EGEB and EUV represent excitation of red green blue andUV hummingbird photoreceptors respectively elicited by aspectrum
x frac14ffiffiffiffiffiffiffiffi
3=4p
EB ffiffiffiffiffiffiffiffiffiffi
1=12p
ethEUV thorn EG thorn ERTHORN
y frac14ffiffiffiffiffiffiffiffi
2=3p
EG ffiffiffiffiffiffiffiffi
1=6p
ethEUV thorn ERTHORN
z frac14ffiffiffiffiffiffiffiffi
1=2p
ethEUV ERTHORN
RStudio was used to make 3D plots of the hummingbird col-our space and ImageJ version 148 (httpimagejnihgovij)was used to create an animation of the Nicotiana flower loci inthe hummingbird colour space Again Nicotiana flower colourloci are close to the origin in the hummingbird colour space sothe graphs presented display only the central portion (either 25or 50 ) of the colour space for clarity To further facilitateinterpretation of these graphs vertices representing individualexcitation of the red green blue and UV photoreceptor typesas well as their excitation vectors from the origin are shownin red green blue and black respectively Other vertices(representing excitation of two or three photoreceptor types) areshown in grey
Cluster analyses
Cluster analyses were used to group spectra based on spectralshape (corresponding to pigmentation) and their position inboth bee and hummingbird colour spaces For spectral colourcategories spectra were normalized to the same integral underthe curve in order to compare combinations of pigments notthe concentration of pigments A distance matrix was calculatedfrom the normalized spectral data in R version 302 (http
wwwR-projectorg) using the dist() function For the bee andhummingbird colour categories the input data were the (x y) or(x y z) coordinates of the spectra in the bee and hummingbirdcolour spaces respectivelyData were first imported into R The function hclust() was
used to perform agglomerative hierarchical clustering based onthe average pairwise distances between groups With this algo-rithm the observed points which are initially all deemed to bedistinct are iteratively assigned to groups until eventually allpoints belong to the same group At each step the average dis-tance between all groups (ie the mean distance from all pointsin group A to all points in group B if either one of these is asingle point then no averaging is needed) is calculated and thetwo groups with the smallest average distance are merged Theorder in which groups are merged can be used to construct adendrogram showing the spatial relationship between all datapoints We can also look at the distribution of merge distancesat each step in the algorithm and use this distribution to esti-mate how many groups are present in the data Points at whichthere is a steep increase in the average between-group distance(lsquoelbowrsquo points) highlight the spatial scale(s) at which there isclustering present in the data By using one of these elbowpoints as a cutoff in the algorithm we can arrive at a distancegrouping that captures the spatial clustering It should be notedthat the determination of where to draw the threshold in a clus-ter analysis is arbitrary but the use of one of these elbow pointsdoes yield meaningful clusters The determination of the spe-cific point from the elbow region to be used to define clusterswas further informed by visual inspection of reflectance spec-tra as well as distributions of colour loci in the perceptual col-our spaces It should be noted that the dendrograms relate tosimilarities in spectral reflectance as well as colour relation-ships perceived by bees and hummingbirds they do not showphylogenetic relationships
Ancestral state reconstruction
To examine the evolution of colour within a phylogeneticcontext ancestral state reconstructions were performed on treesinferred from plastid sequence data Only species for which flo-ral character data are available were included in these analysesBecause polyploid and homoploid hybrids originate via reticu-late evolutionary processes and therefore lack a history oftree-like evolution ancestral characters were reconstructed us-ing only non-hybrid diploid species The states observed inhybrid species were then compared with the ancestral state re-constructions Since sections Repandae and Suaveolentes havediversified to form several species following polyploidizationcharacters were reconstructed for these sections separately toexamine colour evolution subsequent to their origin Fornon-hybrid diploid species individual gene trees yieldsome conflicting topologies nevertheless key nodes for thepurposes of interpreting character evolution in hybrids arerecovered in multiple gene trees and are supported bysupernetwork analyses (Kelly et al 2010) Therefore plastiddata from previously published studies are suitable for theseanalysesPreviously published sequences (Clarkson et al 2004) from
four plastid regions (matK ndhF trnL-F and trnS-G) were
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1121
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
aligned separately using PRANKthornF (Loytynoja and Goldman2008) and then concatenated to create a combined plastid data-set before further optimization by eye in Mesquite version 274(Maddison and Maddison 2008) For Nicotiana attenuata weused GenBank accessions AB040009 and AY098697 for thematK and trnL-F regions respectively (due to likely misidenti-fication of N attenuata material used in Clarkson et al 2004see Clarkson et al 2010) the other two regions were scored asmissing data Phylogenetic reconstruction by Bayesian infer-ence was performed as described in Kelly et al (2013) with theexception that BayesTrees v13 (wwwevolutionreadingacukBayesTreeshtml) was used to construct 95 majority ruleconsensus trees For sections Repandae and Suaveolentes se-quences representing their putative maternal progenitors wereincluded during Bayesian inference to allow rooting of trees butwere pruned from the trees prior to ancestral statereconstructionAncestral states for spectral reflectance colour categories and
presenceabsence of chlorophyll in petals (data inSupplementary Data Table S2) were reconstructed using theparsimony reconstruction method in Mesquite version 274 un-der the unordered states assumption To account for topologicaluncertainty character states were reconstructed over all 36 000post-burn-in trees using the Trace Character Over Trees optionand summarized on the 95 majority rule consensus tree fromthe Bayesian analysis Ancestral states were not calculated forbee or hummingbird colour categories because these are per-ceptual systems and the same colour category can result fromdifferent combinations of pigments (eg both human pink andhuman white flowers which are positive and negative respec-tively for anthocyanin pigmentation are both classified as beebluendashgreen) thus a single colour category does not necessarilyhave a shared evolutionary history
Estimating expected polyploid and homoploid hybrid floralcolour
Polyploid and homoploid hybrid floral colours for each ac-cession were compared with those of their diploid progenitorsfor the spectral bee and hummingbird colour categories definedby cluster analyses Floral colour was classified as lsquoexpectedrsquo ifit fell in the colour category of at least one progenitor or lsquounex-pectedrsquo if it was different from both progenitors Polyploid andhomoploid hybrids were also compared with their diploid pro-genitors for the presence or absence of chlorophyll in corollatissue Chlorophyll absorbs at 675 nm in vivo (Haardt andMaske 1987) therefore the presence of chlorophyll can be in-ferred from reflectance spectra if there is a reflectance mini-mum at 675 nm The presence of chlorophyll in petal tissueappears to be dominant (Brieger 1935 Ohmiya et al 2014)Thus hybrids were classified as expected if they showed chlo-rophyll pigmentation and as unexpected if they did not becauseat least one progenitor possessed chlorophyll in all diploid pro-genitor combinations For natural homoploid hybrids and poly-ploid section Suaveolentes where progenitors can only bedefined to Nicotiana section level comparisons were madewith reconstructed ancestral characters thus only spectral col-our categories and the presenceabsence of chlorophyll wereexamined
Phylogenetic signal in floral traits
In order to statistically test for phylogenetic signal in the phe-notypic trait data (spectral reflectance bee and hummingbirdcolour perception) we used Mantel tests to examine the corre-lation between phylogenetic distance and each of the respectivecontinuous multidimensional traits (eg Cubo et al 2005Muchhala et al 2014) Analyses were restricted to diploid spe-cies excluding homoploid and polyploid hybrids Trees wereedited in Newick format to include additional tips with zerobranch lengths for taxa that are multiple in the trait datasetseither due to colour polymorphism (N otophora) or multipleaccessions (N sylvestris and N obtusifolia var obtusifolia)Statistical analyses were performed in R version 310
Phenotypic distance matrices were first calculated for the threetrait datasets using Euclidean distance and phylogenetic dis-tance matrices were calculated (1) as genetic distance from theplastid alignment and (2) for each of 36 000 post-burn-inBayesian trees using copheneticphylo() part of the APE pack-age version 31-2 (Paradis et al 2004) The second Bayesianset of tests was performed in order to account for evolutionaryprocesses such as saturation and to estimate how phylogeneticuncertainty affects the correlation Mantel tests were performedusing Pearsonrsquos product-moment correlation coefficient with10 000 permutations of each distance matrix to test for signifi-cance the mean P value and its standard deviation were calcu-lated for each set of 36 000 Mantel tests from the Bayesiantrees along with the percentage of trees that gave significantcorrelations The function mantel() from the vegan packagewas used (Oksanen et al 2013)
RESULTS
Petal cell area
Petal cell area was measured to determine whether an increasein ploidy results in larger floral cells Polyploid petal cell areawas significantly larger than in both progenitors in N rustica(ANOVA Ffrac14 371 dffrac14 3 Plt 2 10ndash16) accessions but wasintermediate between progenitors in section Repandae poly-ploids (ANOVA Ffrac14 2492 dffrac14 4 Plt 2 10ndash16 Fig 2)Accessions that were significantly different in cell area (withinpolyploid sections and their progenitors) are represented bydifferent letters above the bars in Fig 2 results from Tukeyrsquoshonest significance tests can be found in Supplementary DataTable S3
Cluster analyses
Nicotiana reflectance spectra were grouped into categoriesbased on spectral shape and position in the bee and humming-bird colour spaces using cluster analyses Bees and humming-birds have different photoreceptor sensitivities and we expectour cluster analyses to reflect these differences in sensoryequipment The analysis based on spectral shape yielded eightcolour categories which roughly corresponded to flowers per-ceived by human observers as magenta red pink UVndashwhitewhite yellow green and dark green (Fig 3) Nicotiana spectraare displayed by spectral colour category in Fig 4A BSupplementary Data Fig S2 The bee colour hexagon clustering
1122 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
resulted in 11 colour categories which fell into the followingareas of bee colour space saturated green UVndashblue high UVUVndashgreen green light green bluendashgreen dark green saturatedUVndashblue saturated UVndashgreen and blue (the last four categorieswere each represented by only a single accessionSupplementary Data Fig S3A) These groups are shown in thebee colour hexagon (Fig 4C) The hummingbird colour spacecluster analysis also produced 11 colour categories saturatedgreen green UVndashwhite UVndashgreen pink white UVndashpinkdark green light pink red and saturated UVndashpink (again thelast four categories include only a single accessionSupplementary Data Fig S3B) These groups are shown in thehummingbird colour space (Fig 4D) and the same graph isprovided as an animation to better display the 3D nature of thecolour space (Supplementary Data Video)
Evolution of spectral reflectance in polyploids and homoploidhybrids
To assess the evolution of polyploid floral colour polyploidand homoploid hybrid accessions were compared with those oftheir progenitors in spectral bee and hummingbird colour cate-gories as well as in the presenceabsence of chlorophyll Thediploid progenitors and approximate age of polyploids andhomoploid hybrids are found in Fig 1 and Supplementary DataTable S4 and the observed and expected floral colours of
polyploid and homoploid hybrids are found in Table 1 Mostpolyploid and homoploid hybrids were similar to at least oneprogenitor in spectral bee and hummingbird colour categoriesbut some fell into unexpected colour categories (Table 1Fig 5 Supplementary Data Fig S4 Supplementary Data FigS5) Over half of the polyploids unexpectedly lacked chloro-phyll (Table 1)
Evolution of colour characters in a phylogenetic context
Reconstructed character states are shown for spectral reflec-tance colour categories (Fig 6) and the presenceabsence ofchlorophyll in petals (Supplementary Data Fig S6) Bee andhummingbird colour categories are also shown for extant spe-cies on the plastid tree (Fig 6) Although the deepest nodeswere largely equivocal evolution of spectral reflectance colourin Nicotiana seemed to be dynamic (Fig 6) Green flowerslikely have three independent origins (1) in sectionsPaniculatae and Undulatae (2) in N langsdorffii and (3) inthe homoploid hybrid N glauca UVndashwhite flowers also seemto have arisen three times independently (1) in sectionTrigonophyllae (2) in N pauciflora and (3) in the homoploidhybrid N linearis Most polyploid and homoploid hybrid spe-cies exhibit a floral colour present in at least one of their pro-genitors However N tabacum 095-55 is red and N glauca isyellow and green unlike their progenitors UVndashwhite flowersseem to have evolved de novo in N linearis UVndashwhite flowersare also found in one of its progenitor sections (in N pauci-flora) but ancestral reconstructions indicate that the floral col-our was most likely white at the ancestral nodes within thesection (Fig 6) This suggests that the evolution of UVndashwhiteflowers in N pauciflora has occurred subsequent to the forma-tion of N linearis and that the two events are likely indepen-dent It is unclear whether UVndashwhite flowers also evolved denovo in N nudicaulis because the ancestral node of sectionRepandae is equivocal The presence of chlorophyll as inferredby light absorption at 675 nm (Haardt and Maske 1987) inNicotiana flowers is ancestral and has been lost three times inN sylvestris N noctiflora and the most recent common ances-tor of N acuminata and N pauciflora (Supplementary DataFig S6)Results from Mantel tests of phylogenetic signal for
Nicotiana floral traits for both genetic distance and the 36 000post-burn-in Bayesian trees are shown in Table 2 All floraltraits were significantly correlated with phylogenetic relation-ships for the Bayesian trees at a significance level of Plt 005Only spectral reflectance was significant for the genetic dis-tance tests whereas bee and hummingbird colour perceptionwere just above the Plt 005 threshold For the Bayesian trees901 662 and 932 of trees were significantly correlatedwith the spectral reflectance bee and hummingbird colour per-ception datasets respectively These results suggest that thesefloral traits are weakly constrained by phylogeny
DISCUSSION
Nicotiana is remarkable in its range of spectral reflectanceflower colours (white UVndashwhite pink magenta red yellowgreen and dark green Fig 3) and in the variety of pollinators
0
200
400
600
800
1000
1200C
ell
are
a (
microm
2)
N rustica
N paniculata
N undulata
N sylvestris
N obtusifolia
N nudicaulis
N repanda
N stocktonii
N rustica var asiatica
N rustica var pavonii
Repandae
a
a
b
b
cc
c c
d
Maternal progenitor
Paternal progenitor
Polyploids
Polyploids
Polyploids
FIG 2 Petal cell area from polyploids and their progenitors Within each poly-ploid group bars with different letters represent significantly different mean cell
areas
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1123
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
that visit the flowers (moth bird bee bat Knapp 2010) Theperception of these spectral colours also changes with visualsystem (bee or hummingbird) Here we describe a complex dy-namic in the evolution of floral colour in Nicotiana Spectral re-flectance and bee and hummingbird colour perception arecorrelated with phylogeny but multiple independent origins ofvarious combinations of pigmentation suggest that the evolutionof floral colour is not entirely phylogenetically constrained
Petal cell size evolution in polyploids
Cell size is expected to increase following polyploidizationdue to the increase in genome size (Beaulieu et al 2008) Thesignificantly larger petal cells of N rustica (lt02 million yearsold myo) and the intermediate cell size of section Repandae(45myo) suggest that polyploids may revert to a diploid-like cell size over time similar to the genome downsizing ob-served in polyploids (Leitch and Bennett 2004) Howeverpetal cell size differences within section Repandae do not seemto be linked to genome size N nudicaulis and N repanda sharesimilar cell sizes but have substantially different genome sizes(Leitch et al 2008) The concentration of pigment in floralcells is controlled by both the amount of pigment present andthe cell size Nicotiana rustica has significantly larger petalcells than its progenitors (close to the sum Fig 2) and displaysan intermediate brightness (the area under the reflectance curvea proxy for pigment concentration) between its progenitors
(Supplementary Data Fig S4D) This is expected if the poly-ploid inherits the sum of both cell size and amount of pigmentpresent from its progenitors
Polyploid divergence in floral colour
Many younger polyploids (lt02myo) display unexpectedfloral colours considering those of their diploid progenitorsNone of the natural and synthetic N tabacum accessions pos-sess chlorophyll which is unexpected given its presence in atleast one progenitor species Nicotiana tabacum 095-55 alsohas unexpected spectral bee and hummingbird colour giventhe colour categories of the progenitor species Similarly syn-thetic N tabacum QM has unexpected colours in bee and hum-mingbird perception Because this accession is synthetic theparents are known and thus its unexpected phenotype can beclassified as transgressive or outside the range of its progeni-tors due solely to polyploidy and hybridization Most N rusticaaccessions have unexpected bee colour (four of these aresynthetic and are therefore also transgressive) and N arentsiihas unexpected hummingbird colour (Table 1 Fig 6) Despitethe divergence of floral spectra associated with polyploidybehavioural studies are needed to determine whether the beeand hummingbird colour categories delineated here actuallyelicit different responses in pollinatorsMost older polyploids (1ndash10myo) are similar in floral col-
our category to at least one of their progenitors N clevelandii
synth
etic
N ru
stic
a P
UE
1-R
1 S
1
N ru
stic
a v
ar a
sia
tica
synth
etic
N ru
stic
a P
UE
1-R
10 S
0
synth
etic
PU
E1 F
1synth
etic
UxP
N b
enavid
esii
N ru
stic
a v
ar p
avonii
N p
anic
ula
taN
langsdorffii
N g
lauca 5
1751 g
reen
N u
ndula
taN
raim
ondii
N k
nig
htia
na
N g
lauca 5
1751 y
ello
w
N g
lauca 5
1725
N p
lum
bagin
ifolia
N g
ossei
N q
uadriv
alv
is T
W18
N o
ccid
enta
lis s
ubsp h
esperis
N
megalo
sip
hon
N s
uaveole
ns
N n
esophila
N s
tockto
nii
N q
uadriv
alv
is 9
04750042
N a
ttenuata
N re
panda
N s
ylv
estris
A04750326
N
benth
am
iana
N m
uta
bilis
CP
G3 w
hite
N
muta
bilis
CP
G12456 w
hite
N
wig
andio
ides
N a
cum
inata
N fo
rste
riN
cle
vela
ndii
N n
octiflo
raN
x o
btu
sia
ta lin
e 2
N
x o
btu
sia
ta lin
e 1
N
mie
rsii
N p
etu
nio
ides
N x
obtu
sia
ta lin
e 5
N
sylv
estris
6898
N
oto
phora
white
N
are
nts
iiT
H3
2N
tabacum
lsquoChulu
manirsquo
N o
btu
sifo
lia v
ar p
alm
eri
N o
btu
sifo
lia v
ar o
btu
sifo
lia lsquoB
ald
win
rsquoN
linearis
TW
77
N n
udic
aulis
N o
btu
sifo
lia v
ar o
btu
sifo
lia T
W143
N lin
earis
964750099
N
pauciflo
raN
tom
ento
sifo
rmis
N o
tophora
pin
k
N g
lutin
osa
synth
etic
N ta
bacum
TH
37
N
tabacum
51789
synth
etic
N ta
bacum
QM
N
tabacum
095-5
5
N s
etc
hellii
N m
uta
bilis
CP
G3 p
ink
N m
uta
bilis
CP
G12456 p
ink
10
8
6
4
2
0
Dis
tan
ce
Spectral categories
MagentaRedPinkUV-white
WhiteYellowDark greenGreen
Dark greenSaturated greenGreenUV-whiteUV-green
PinkLight pinkWhiteUV-pinkRedSaturated UV-pink
SpecBeeHum
SpectralDark greenSaturated greenSaturated UV-blueUV-blueSaturated UV-greenHigh UV
UV-greenGreenLight greenBlueBlue-green
Bee Hummingbird
FIG 3 Dendrograms based on distance cluster analyses for spectral reflectance Coloured circles on the dendrogram represent distinct colour categories as determinedby the chosen threshold (dashed line) Dendrograms were similarly constructed for bee and hummingbird colour (see Supplementary Data Fig S3) The lines of col-oured circles at the tips of the dendrograms signify the category each taxon is assigned to in spectral bee and hummingbird colour as labelled (Spec Bee and Hum)for comparison between spectral categories and those of different visual systems Diploids polyploids and homoploids are denoted by black blue and orange text
respectively
1124 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
0
0middot2
0middot4
0middot6
0middot8
1middot0
Reflecta
nce
Wavelength (nm)
0
0middot2
0middot4
0middot6
0middot8
1middot0
300 350 400 450 500 550 600 650 700 300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
N benavidesii
N glauca 51751 g
N langsdorffii
N paniculata
N raimondii
N rustica var asiatica
N rustica var pavonii
synthetic UxP
synthetic F1
syn N rustica S0
syn N rustica S1
N undulata
N glutinosa
N otophora p
N tabacum 51789
syn N tabacum QM
syn N tabacum TH37
N tomentosiformis
A B
G 50
R 50
B 50
UV 50
Hummingbird colour space
White
0middot40middot4
ndash0middot40middot4
0middot4
0middot4
y
z
UV-white
Pink
Synthetic N tabacum QM
UV-pink
N tabacum 095-55
N setchellii
Green
N langsdorffii
Saturated green
UV-green
C DBee colour hexagon
Blue-green
Light green
Green
Dark green
Sat green
UV-green
Sat UV-green
B
B-G
G
UV-G
UV
UV-B
1 acum2 aren3 atten4 bena5 benth6 clev7 forst8 glau259 glau51y10 glau51g11 glut12 goss13 knight14 langs15 lin964716 linTW7717 mega18 mier19 mut1w20 mut1p21 mut3w22mut3p23 neso24 noct25 nudi26 xobtus127 xobtus228 xobtus529 obtusB30 obtusTW31 obtuspalm
32 occhesp33 otoph w34 otoph p35 pani36 pauc37 petun38 plumba39 quad904740 quadTW1841 raim42 repa43 rustasi44 rustpav45 syn UxP46 syn F147 syn rust S048 syn rust S149 setch50 stock51 suav52 sylv689853 sylvA04754 tab0955555 tab5178956 tabchulu57 syntabQM58 syntabTH3759 tomtform60 undu61 wigan62TH32
High UV
UV-blue
Sat UV-blue
Blue
FIG 4 (A B) Nicotiana reflectance spectra from 300 to 700nm which roughly correspond to colours perceived by human observers as pink (A) and green (B) SeeSupplementary Data Fig S2 for other spectral colour categories Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploidhybrid taxa Abbreviations p pink syn synthetic g green (C) Colour hexagon displaying the distribution of Nicotiana colour loci in bee colour space The hexa-gon has been scaled so that vertices represent 40 excitation of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-GUVndashgreen Bee colour categories are delineated by coloured ovals sat saturated Nicotiana species abbreviations are as follows acum acuminata aren arentsiiatten attenuata benavid benavidesii benth benthamiana clev clevelandii forst forsteri glau25 glauca 51725 glau51y glauca 51751 yellow glau51g glauca51751 green glut glutinosa goss gossei knight knightiana langs langsdorffii lin9647 linearis 964750099 linTW77 linearis TW77 mega megalosiphon miermiersii mut1w mutabilis CPG12456 white mut1p mutabilis CPG12456 pink mut3w mutabilis CPG3 white mut3p mutabilis CPG3 pink neso nesophila noctnoctiflora nudi nudicaulis obtus1 obtusiata line 1 obtus2 obtusiata line 2 obtus5 obtusiata line 5 obtusB obtusifolia var obtusifolia lsquoBaldwinrsquoobtusTW obtusifolia var obtusifolia TW143 obtuspalm obtusifolia var palmeri occhesp occidentalis subsp hesperis otoph w otophora white otoph p otophorapink pani paniculata pauc pauciflora petun petunioides plumba plumbaginifolia quad9047 quadrivalvis 904750042 quadTW18 quadrivalvis TW18 raimraimondii repa repanda rustasi rustica var asiatica rustpav rustica var pavonii syn UP synthetic UP syn F1 synthetic PUE1 F1 synrust S0 synthetic rus-tica PUE1-R10 S0 synrust S1 synthetic rustica PUE1-R1 S1 setch setchellii stock stocktonii suav suaveolens sylv6898 sylvestris 6898 sylvA047 sylvestrisA04750326 tab09555 tabacum 095-55 tab51789 tabacum 51789 tabchulu tabacum lsquoChulumanirsquo syntabQM synthetic tabacum QM syntabTH37 synthetictabacum TH37 tomtform tomentosiformis undu undulata wigan wigandioides TH32 TH32 synthetic N sylvestrisN otophora polyploid (D) Distribution ofNicotiana spectral loci in hummingbird colour space Vertices of the hummingbird colour space represent 50 excitation of the photoreceptors single photoreceptortype vertices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows repre-sent the vectors of these photoreceptors from the origin of the hummingbird colour space Nicotiana loci are coloured according to hummingbird colour categories(Supplementary Data Fig S3B) but are labelled with the accession name if the category includes only one taxon See Supplementary Data Video for an animation
of this graph
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1125
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
is the exception because it falls into an unexpected humming-bird colour category given the progenitor species (Fig 6)Section Repandae polyploids seem to have evolved to be eitherlike their maternal (Nicotiana nesophila N repanda andN stocktonii) or paternal (N nudicaulis) progenitor (Fig 6)The maternal progenitor N sylvestris is no longer sympatricwith any of the section Repandae polyploids therefore Nnesophila N repanda and N stocktonii can occupy the samepollination niche as their maternal progenitor without competi-tion Similarly section Suaveolentes is native to Australasiaexcept for one species in Namibia Africa and is not sympatricwith its progenitor sections in South America (Goodspeed1954) these polyploids and their diploid progenitors displaysimilar floral colours except N pauciflora which evolvedspectrally UVndashwhite flowers after the formation of sectionSuaveolentes (Fig 6) It is possible that there is less competitionfor pollinators and therefore reduced selective pressure to-wards floral colour diversification when polyploid species arenot sympatric with their diploid progenitors as is seen in
Iochrominae (Solanaceae) which have a broader range of floralcolours when species are sympatric (Muchhala et al 2014)However floral colour evolution can also be driven by geneticdrift or selection on pleiotropic effects of floral genes (Chittkaet al 2001 Rausher 2008) Furthermore anthocyanins andflavonoids are important for UV protection can defend againstfungi act as signal molecules and play a role in male fertility insome species in addition to their roles in signalling to pollina-tors (Shirley 1996)Over half of polyploids have an unexpected inheritance pat-
tern for chlorophyll if it is assumed that the presence of chloro-phyll is a dominant character Those polyploids that deviatefrom expectation span an age range of synthetic to 10myoand always lack chlorophyll in their petals which is in linewith the direction of the shifts observed in the divergence ofdiploid species (Supplementary Data Fig S6) In carnation thedifference in chlorophyll concentration between white andgreen flowers is likely caused by downregulation of chlorophyllbiosynthesis genes in white flowers genes involved in
TABLE 1 Polyploid and homoploid hybrid observed and expected floral colours
Species Spectral Bee Hummingbird Chlorophyll
Observed Expected Observed Expected Observed Expected Observed Expected
N tabacum 095-55 R W P UV-B B-G sUV-P W P N CN tabacum 51789 P W P B-G B-G P W P N CN tabacum lsquoChulumanirsquo W W P B-G B-G P W P N Csyn N tabacum QM P W P B B-G LP W P N Csyn N tabacum TH37 P W P B-G B-G P W P N CTH32 P W P B-G B-G P W P N CN rustica var asiatica G G LG G B-G G G W C CN rustica var pavonii G G G G B-G G G W C Csyn UP (homoploid) G G UV-G G B-G UV-G G W C Csyn F1 (homoploid) G G LG G B-G G G W C Csyn N rustica S0 G G LG G B-G G G W C Csyn N rustica S1 G G LG G B-G G G W C CN arentsii W G W B-G B-G P W C CN clevelandii W UV-W W B-G UV B-G P UV-W W N CN quadrivalvis TW18 W UV-W W B-G UV B-G W UV-W W C CN quadrivalvis 9047 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 1 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 2 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 5 W UV-W W B-G UV B-G W UV-W W C CN repanda W W UV-W B-G B-G UV W W UV-W N CN nesophila W W UV-W B-G B-G UV W W UV-W N CN stocktonii W W UV-W B-G B-G UV W W UV-W C CN nudicaulis UV-W W UV-W UV B-G UV UV-W W UV-W C CN benthamiana W W ndash ndash ndash ndash N CN forsteri W W ndash ndash ndash ndash C CN gossei W W ndash ndash ndash ndash N CN megalosiphon W W ndash ndash ndash ndash N CN occidentalis W W ndash ndash ndash ndash N CN suaveolens W W ndash ndash ndash ndash N C
N glauca 51725 Y W ndash ndash ndash ndash C CN glauca 51751 YG W ndash ndash ndash ndash C CN linearis TW77 UV-W W ndash ndash ndash ndash C CN linearis 9647 UV-W W ndash ndash ndash ndash C CN glutinosa P P W G ndash ndash ndash ndash C C
The top block consists of polyploid accessions and the bottom block includes homoploid hybridsItalic denotes an unexpected phenotype given the colour categories of the progenitorsProgenitor bee and hummingbird colour categories are unknown for section Suaveolentes and natural homoploid hybrids (see text)R red W white P pink G green UV-W UVndashwhite Y yellow UV-B UVndashblue B-G bluendashgreen B blue LG light green UV high UV UV-G
UVndashgreen sUV-P saturated UVndashpink LP light pink N no chlorophyll C chlorophyll syn synthetic 9047 represents 904750042 9647 represents964750099
1126 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
aligned separately using PRANKthornF (Loytynoja and Goldman2008) and then concatenated to create a combined plastid data-set before further optimization by eye in Mesquite version 274(Maddison and Maddison 2008) For Nicotiana attenuata weused GenBank accessions AB040009 and AY098697 for thematK and trnL-F regions respectively (due to likely misidenti-fication of N attenuata material used in Clarkson et al 2004see Clarkson et al 2010) the other two regions were scored asmissing data Phylogenetic reconstruction by Bayesian infer-ence was performed as described in Kelly et al (2013) with theexception that BayesTrees v13 (wwwevolutionreadingacukBayesTreeshtml) was used to construct 95 majority ruleconsensus trees For sections Repandae and Suaveolentes se-quences representing their putative maternal progenitors wereincluded during Bayesian inference to allow rooting of trees butwere pruned from the trees prior to ancestral statereconstructionAncestral states for spectral reflectance colour categories and
presenceabsence of chlorophyll in petals (data inSupplementary Data Table S2) were reconstructed using theparsimony reconstruction method in Mesquite version 274 un-der the unordered states assumption To account for topologicaluncertainty character states were reconstructed over all 36 000post-burn-in trees using the Trace Character Over Trees optionand summarized on the 95 majority rule consensus tree fromthe Bayesian analysis Ancestral states were not calculated forbee or hummingbird colour categories because these are per-ceptual systems and the same colour category can result fromdifferent combinations of pigments (eg both human pink andhuman white flowers which are positive and negative respec-tively for anthocyanin pigmentation are both classified as beebluendashgreen) thus a single colour category does not necessarilyhave a shared evolutionary history
Estimating expected polyploid and homoploid hybrid floralcolour
Polyploid and homoploid hybrid floral colours for each ac-cession were compared with those of their diploid progenitorsfor the spectral bee and hummingbird colour categories definedby cluster analyses Floral colour was classified as lsquoexpectedrsquo ifit fell in the colour category of at least one progenitor or lsquounex-pectedrsquo if it was different from both progenitors Polyploid andhomoploid hybrids were also compared with their diploid pro-genitors for the presence or absence of chlorophyll in corollatissue Chlorophyll absorbs at 675 nm in vivo (Haardt andMaske 1987) therefore the presence of chlorophyll can be in-ferred from reflectance spectra if there is a reflectance mini-mum at 675 nm The presence of chlorophyll in petal tissueappears to be dominant (Brieger 1935 Ohmiya et al 2014)Thus hybrids were classified as expected if they showed chlo-rophyll pigmentation and as unexpected if they did not becauseat least one progenitor possessed chlorophyll in all diploid pro-genitor combinations For natural homoploid hybrids and poly-ploid section Suaveolentes where progenitors can only bedefined to Nicotiana section level comparisons were madewith reconstructed ancestral characters thus only spectral col-our categories and the presenceabsence of chlorophyll wereexamined
Phylogenetic signal in floral traits
In order to statistically test for phylogenetic signal in the phe-notypic trait data (spectral reflectance bee and hummingbirdcolour perception) we used Mantel tests to examine the corre-lation between phylogenetic distance and each of the respectivecontinuous multidimensional traits (eg Cubo et al 2005Muchhala et al 2014) Analyses were restricted to diploid spe-cies excluding homoploid and polyploid hybrids Trees wereedited in Newick format to include additional tips with zerobranch lengths for taxa that are multiple in the trait datasetseither due to colour polymorphism (N otophora) or multipleaccessions (N sylvestris and N obtusifolia var obtusifolia)Statistical analyses were performed in R version 310
Phenotypic distance matrices were first calculated for the threetrait datasets using Euclidean distance and phylogenetic dis-tance matrices were calculated (1) as genetic distance from theplastid alignment and (2) for each of 36 000 post-burn-inBayesian trees using copheneticphylo() part of the APE pack-age version 31-2 (Paradis et al 2004) The second Bayesianset of tests was performed in order to account for evolutionaryprocesses such as saturation and to estimate how phylogeneticuncertainty affects the correlation Mantel tests were performedusing Pearsonrsquos product-moment correlation coefficient with10 000 permutations of each distance matrix to test for signifi-cance the mean P value and its standard deviation were calcu-lated for each set of 36 000 Mantel tests from the Bayesiantrees along with the percentage of trees that gave significantcorrelations The function mantel() from the vegan packagewas used (Oksanen et al 2013)
RESULTS
Petal cell area
Petal cell area was measured to determine whether an increasein ploidy results in larger floral cells Polyploid petal cell areawas significantly larger than in both progenitors in N rustica(ANOVA Ffrac14 371 dffrac14 3 Plt 2 10ndash16) accessions but wasintermediate between progenitors in section Repandae poly-ploids (ANOVA Ffrac14 2492 dffrac14 4 Plt 2 10ndash16 Fig 2)Accessions that were significantly different in cell area (withinpolyploid sections and their progenitors) are represented bydifferent letters above the bars in Fig 2 results from Tukeyrsquoshonest significance tests can be found in Supplementary DataTable S3
Cluster analyses
Nicotiana reflectance spectra were grouped into categoriesbased on spectral shape and position in the bee and humming-bird colour spaces using cluster analyses Bees and humming-birds have different photoreceptor sensitivities and we expectour cluster analyses to reflect these differences in sensoryequipment The analysis based on spectral shape yielded eightcolour categories which roughly corresponded to flowers per-ceived by human observers as magenta red pink UVndashwhitewhite yellow green and dark green (Fig 3) Nicotiana spectraare displayed by spectral colour category in Fig 4A BSupplementary Data Fig S2 The bee colour hexagon clustering
1122 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
resulted in 11 colour categories which fell into the followingareas of bee colour space saturated green UVndashblue high UVUVndashgreen green light green bluendashgreen dark green saturatedUVndashblue saturated UVndashgreen and blue (the last four categorieswere each represented by only a single accessionSupplementary Data Fig S3A) These groups are shown in thebee colour hexagon (Fig 4C) The hummingbird colour spacecluster analysis also produced 11 colour categories saturatedgreen green UVndashwhite UVndashgreen pink white UVndashpinkdark green light pink red and saturated UVndashpink (again thelast four categories include only a single accessionSupplementary Data Fig S3B) These groups are shown in thehummingbird colour space (Fig 4D) and the same graph isprovided as an animation to better display the 3D nature of thecolour space (Supplementary Data Video)
Evolution of spectral reflectance in polyploids and homoploidhybrids
To assess the evolution of polyploid floral colour polyploidand homoploid hybrid accessions were compared with those oftheir progenitors in spectral bee and hummingbird colour cate-gories as well as in the presenceabsence of chlorophyll Thediploid progenitors and approximate age of polyploids andhomoploid hybrids are found in Fig 1 and Supplementary DataTable S4 and the observed and expected floral colours of
polyploid and homoploid hybrids are found in Table 1 Mostpolyploid and homoploid hybrids were similar to at least oneprogenitor in spectral bee and hummingbird colour categoriesbut some fell into unexpected colour categories (Table 1Fig 5 Supplementary Data Fig S4 Supplementary Data FigS5) Over half of the polyploids unexpectedly lacked chloro-phyll (Table 1)
Evolution of colour characters in a phylogenetic context
Reconstructed character states are shown for spectral reflec-tance colour categories (Fig 6) and the presenceabsence ofchlorophyll in petals (Supplementary Data Fig S6) Bee andhummingbird colour categories are also shown for extant spe-cies on the plastid tree (Fig 6) Although the deepest nodeswere largely equivocal evolution of spectral reflectance colourin Nicotiana seemed to be dynamic (Fig 6) Green flowerslikely have three independent origins (1) in sectionsPaniculatae and Undulatae (2) in N langsdorffii and (3) inthe homoploid hybrid N glauca UVndashwhite flowers also seemto have arisen three times independently (1) in sectionTrigonophyllae (2) in N pauciflora and (3) in the homoploidhybrid N linearis Most polyploid and homoploid hybrid spe-cies exhibit a floral colour present in at least one of their pro-genitors However N tabacum 095-55 is red and N glauca isyellow and green unlike their progenitors UVndashwhite flowersseem to have evolved de novo in N linearis UVndashwhite flowersare also found in one of its progenitor sections (in N pauci-flora) but ancestral reconstructions indicate that the floral col-our was most likely white at the ancestral nodes within thesection (Fig 6) This suggests that the evolution of UVndashwhiteflowers in N pauciflora has occurred subsequent to the forma-tion of N linearis and that the two events are likely indepen-dent It is unclear whether UVndashwhite flowers also evolved denovo in N nudicaulis because the ancestral node of sectionRepandae is equivocal The presence of chlorophyll as inferredby light absorption at 675 nm (Haardt and Maske 1987) inNicotiana flowers is ancestral and has been lost three times inN sylvestris N noctiflora and the most recent common ances-tor of N acuminata and N pauciflora (Supplementary DataFig S6)Results from Mantel tests of phylogenetic signal for
Nicotiana floral traits for both genetic distance and the 36 000post-burn-in Bayesian trees are shown in Table 2 All floraltraits were significantly correlated with phylogenetic relation-ships for the Bayesian trees at a significance level of Plt 005Only spectral reflectance was significant for the genetic dis-tance tests whereas bee and hummingbird colour perceptionwere just above the Plt 005 threshold For the Bayesian trees901 662 and 932 of trees were significantly correlatedwith the spectral reflectance bee and hummingbird colour per-ception datasets respectively These results suggest that thesefloral traits are weakly constrained by phylogeny
DISCUSSION
Nicotiana is remarkable in its range of spectral reflectanceflower colours (white UVndashwhite pink magenta red yellowgreen and dark green Fig 3) and in the variety of pollinators
0
200
400
600
800
1000
1200C
ell
are
a (
microm
2)
N rustica
N paniculata
N undulata
N sylvestris
N obtusifolia
N nudicaulis
N repanda
N stocktonii
N rustica var asiatica
N rustica var pavonii
Repandae
a
a
b
b
cc
c c
d
Maternal progenitor
Paternal progenitor
Polyploids
Polyploids
Polyploids
FIG 2 Petal cell area from polyploids and their progenitors Within each poly-ploid group bars with different letters represent significantly different mean cell
areas
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1123
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
that visit the flowers (moth bird bee bat Knapp 2010) Theperception of these spectral colours also changes with visualsystem (bee or hummingbird) Here we describe a complex dy-namic in the evolution of floral colour in Nicotiana Spectral re-flectance and bee and hummingbird colour perception arecorrelated with phylogeny but multiple independent origins ofvarious combinations of pigmentation suggest that the evolutionof floral colour is not entirely phylogenetically constrained
Petal cell size evolution in polyploids
Cell size is expected to increase following polyploidizationdue to the increase in genome size (Beaulieu et al 2008) Thesignificantly larger petal cells of N rustica (lt02 million yearsold myo) and the intermediate cell size of section Repandae(45myo) suggest that polyploids may revert to a diploid-like cell size over time similar to the genome downsizing ob-served in polyploids (Leitch and Bennett 2004) Howeverpetal cell size differences within section Repandae do not seemto be linked to genome size N nudicaulis and N repanda sharesimilar cell sizes but have substantially different genome sizes(Leitch et al 2008) The concentration of pigment in floralcells is controlled by both the amount of pigment present andthe cell size Nicotiana rustica has significantly larger petalcells than its progenitors (close to the sum Fig 2) and displaysan intermediate brightness (the area under the reflectance curvea proxy for pigment concentration) between its progenitors
(Supplementary Data Fig S4D) This is expected if the poly-ploid inherits the sum of both cell size and amount of pigmentpresent from its progenitors
Polyploid divergence in floral colour
Many younger polyploids (lt02myo) display unexpectedfloral colours considering those of their diploid progenitorsNone of the natural and synthetic N tabacum accessions pos-sess chlorophyll which is unexpected given its presence in atleast one progenitor species Nicotiana tabacum 095-55 alsohas unexpected spectral bee and hummingbird colour giventhe colour categories of the progenitor species Similarly syn-thetic N tabacum QM has unexpected colours in bee and hum-mingbird perception Because this accession is synthetic theparents are known and thus its unexpected phenotype can beclassified as transgressive or outside the range of its progeni-tors due solely to polyploidy and hybridization Most N rusticaaccessions have unexpected bee colour (four of these aresynthetic and are therefore also transgressive) and N arentsiihas unexpected hummingbird colour (Table 1 Fig 6) Despitethe divergence of floral spectra associated with polyploidybehavioural studies are needed to determine whether the beeand hummingbird colour categories delineated here actuallyelicit different responses in pollinatorsMost older polyploids (1ndash10myo) are similar in floral col-
our category to at least one of their progenitors N clevelandii
synth
etic
N ru
stic
a P
UE
1-R
1 S
1
N ru
stic
a v
ar a
sia
tica
synth
etic
N ru
stic
a P
UE
1-R
10 S
0
synth
etic
PU
E1 F
1synth
etic
UxP
N b
enavid
esii
N ru
stic
a v
ar p
avonii
N p
anic
ula
taN
langsdorffii
N g
lauca 5
1751 g
reen
N u
ndula
taN
raim
ondii
N k
nig
htia
na
N g
lauca 5
1751 y
ello
w
N g
lauca 5
1725
N p
lum
bagin
ifolia
N g
ossei
N q
uadriv
alv
is T
W18
N o
ccid
enta
lis s
ubsp h
esperis
N
megalo
sip
hon
N s
uaveole
ns
N n
esophila
N s
tockto
nii
N q
uadriv
alv
is 9
04750042
N a
ttenuata
N re
panda
N s
ylv
estris
A04750326
N
benth
am
iana
N m
uta
bilis
CP
G3 w
hite
N
muta
bilis
CP
G12456 w
hite
N
wig
andio
ides
N a
cum
inata
N fo
rste
riN
cle
vela
ndii
N n
octiflo
raN
x o
btu
sia
ta lin
e 2
N
x o
btu
sia
ta lin
e 1
N
mie
rsii
N p
etu
nio
ides
N x
obtu
sia
ta lin
e 5
N
sylv
estris
6898
N
oto
phora
white
N
are
nts
iiT
H3
2N
tabacum
lsquoChulu
manirsquo
N o
btu
sifo
lia v
ar p
alm
eri
N o
btu
sifo
lia v
ar o
btu
sifo
lia lsquoB
ald
win
rsquoN
linearis
TW
77
N n
udic
aulis
N o
btu
sifo
lia v
ar o
btu
sifo
lia T
W143
N lin
earis
964750099
N
pauciflo
raN
tom
ento
sifo
rmis
N o
tophora
pin
k
N g
lutin
osa
synth
etic
N ta
bacum
TH
37
N
tabacum
51789
synth
etic
N ta
bacum
QM
N
tabacum
095-5
5
N s
etc
hellii
N m
uta
bilis
CP
G3 p
ink
N m
uta
bilis
CP
G12456 p
ink
10
8
6
4
2
0
Dis
tan
ce
Spectral categories
MagentaRedPinkUV-white
WhiteYellowDark greenGreen
Dark greenSaturated greenGreenUV-whiteUV-green
PinkLight pinkWhiteUV-pinkRedSaturated UV-pink
SpecBeeHum
SpectralDark greenSaturated greenSaturated UV-blueUV-blueSaturated UV-greenHigh UV
UV-greenGreenLight greenBlueBlue-green
Bee Hummingbird
FIG 3 Dendrograms based on distance cluster analyses for spectral reflectance Coloured circles on the dendrogram represent distinct colour categories as determinedby the chosen threshold (dashed line) Dendrograms were similarly constructed for bee and hummingbird colour (see Supplementary Data Fig S3) The lines of col-oured circles at the tips of the dendrograms signify the category each taxon is assigned to in spectral bee and hummingbird colour as labelled (Spec Bee and Hum)for comparison between spectral categories and those of different visual systems Diploids polyploids and homoploids are denoted by black blue and orange text
respectively
1124 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
0
0middot2
0middot4
0middot6
0middot8
1middot0
Reflecta
nce
Wavelength (nm)
0
0middot2
0middot4
0middot6
0middot8
1middot0
300 350 400 450 500 550 600 650 700 300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
N benavidesii
N glauca 51751 g
N langsdorffii
N paniculata
N raimondii
N rustica var asiatica
N rustica var pavonii
synthetic UxP
synthetic F1
syn N rustica S0
syn N rustica S1
N undulata
N glutinosa
N otophora p
N tabacum 51789
syn N tabacum QM
syn N tabacum TH37
N tomentosiformis
A B
G 50
R 50
B 50
UV 50
Hummingbird colour space
White
0middot40middot4
ndash0middot40middot4
0middot4
0middot4
y
z
UV-white
Pink
Synthetic N tabacum QM
UV-pink
N tabacum 095-55
N setchellii
Green
N langsdorffii
Saturated green
UV-green
C DBee colour hexagon
Blue-green
Light green
Green
Dark green
Sat green
UV-green
Sat UV-green
B
B-G
G
UV-G
UV
UV-B
1 acum2 aren3 atten4 bena5 benth6 clev7 forst8 glau259 glau51y10 glau51g11 glut12 goss13 knight14 langs15 lin964716 linTW7717 mega18 mier19 mut1w20 mut1p21 mut3w22mut3p23 neso24 noct25 nudi26 xobtus127 xobtus228 xobtus529 obtusB30 obtusTW31 obtuspalm
32 occhesp33 otoph w34 otoph p35 pani36 pauc37 petun38 plumba39 quad904740 quadTW1841 raim42 repa43 rustasi44 rustpav45 syn UxP46 syn F147 syn rust S048 syn rust S149 setch50 stock51 suav52 sylv689853 sylvA04754 tab0955555 tab5178956 tabchulu57 syntabQM58 syntabTH3759 tomtform60 undu61 wigan62TH32
High UV
UV-blue
Sat UV-blue
Blue
FIG 4 (A B) Nicotiana reflectance spectra from 300 to 700nm which roughly correspond to colours perceived by human observers as pink (A) and green (B) SeeSupplementary Data Fig S2 for other spectral colour categories Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploidhybrid taxa Abbreviations p pink syn synthetic g green (C) Colour hexagon displaying the distribution of Nicotiana colour loci in bee colour space The hexa-gon has been scaled so that vertices represent 40 excitation of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-GUVndashgreen Bee colour categories are delineated by coloured ovals sat saturated Nicotiana species abbreviations are as follows acum acuminata aren arentsiiatten attenuata benavid benavidesii benth benthamiana clev clevelandii forst forsteri glau25 glauca 51725 glau51y glauca 51751 yellow glau51g glauca51751 green glut glutinosa goss gossei knight knightiana langs langsdorffii lin9647 linearis 964750099 linTW77 linearis TW77 mega megalosiphon miermiersii mut1w mutabilis CPG12456 white mut1p mutabilis CPG12456 pink mut3w mutabilis CPG3 white mut3p mutabilis CPG3 pink neso nesophila noctnoctiflora nudi nudicaulis obtus1 obtusiata line 1 obtus2 obtusiata line 2 obtus5 obtusiata line 5 obtusB obtusifolia var obtusifolia lsquoBaldwinrsquoobtusTW obtusifolia var obtusifolia TW143 obtuspalm obtusifolia var palmeri occhesp occidentalis subsp hesperis otoph w otophora white otoph p otophorapink pani paniculata pauc pauciflora petun petunioides plumba plumbaginifolia quad9047 quadrivalvis 904750042 quadTW18 quadrivalvis TW18 raimraimondii repa repanda rustasi rustica var asiatica rustpav rustica var pavonii syn UP synthetic UP syn F1 synthetic PUE1 F1 synrust S0 synthetic rus-tica PUE1-R10 S0 synrust S1 synthetic rustica PUE1-R1 S1 setch setchellii stock stocktonii suav suaveolens sylv6898 sylvestris 6898 sylvA047 sylvestrisA04750326 tab09555 tabacum 095-55 tab51789 tabacum 51789 tabchulu tabacum lsquoChulumanirsquo syntabQM synthetic tabacum QM syntabTH37 synthetictabacum TH37 tomtform tomentosiformis undu undulata wigan wigandioides TH32 TH32 synthetic N sylvestrisN otophora polyploid (D) Distribution ofNicotiana spectral loci in hummingbird colour space Vertices of the hummingbird colour space represent 50 excitation of the photoreceptors single photoreceptortype vertices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows repre-sent the vectors of these photoreceptors from the origin of the hummingbird colour space Nicotiana loci are coloured according to hummingbird colour categories(Supplementary Data Fig S3B) but are labelled with the accession name if the category includes only one taxon See Supplementary Data Video for an animation
of this graph
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1125
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
is the exception because it falls into an unexpected humming-bird colour category given the progenitor species (Fig 6)Section Repandae polyploids seem to have evolved to be eitherlike their maternal (Nicotiana nesophila N repanda andN stocktonii) or paternal (N nudicaulis) progenitor (Fig 6)The maternal progenitor N sylvestris is no longer sympatricwith any of the section Repandae polyploids therefore Nnesophila N repanda and N stocktonii can occupy the samepollination niche as their maternal progenitor without competi-tion Similarly section Suaveolentes is native to Australasiaexcept for one species in Namibia Africa and is not sympatricwith its progenitor sections in South America (Goodspeed1954) these polyploids and their diploid progenitors displaysimilar floral colours except N pauciflora which evolvedspectrally UVndashwhite flowers after the formation of sectionSuaveolentes (Fig 6) It is possible that there is less competitionfor pollinators and therefore reduced selective pressure to-wards floral colour diversification when polyploid species arenot sympatric with their diploid progenitors as is seen in
Iochrominae (Solanaceae) which have a broader range of floralcolours when species are sympatric (Muchhala et al 2014)However floral colour evolution can also be driven by geneticdrift or selection on pleiotropic effects of floral genes (Chittkaet al 2001 Rausher 2008) Furthermore anthocyanins andflavonoids are important for UV protection can defend againstfungi act as signal molecules and play a role in male fertility insome species in addition to their roles in signalling to pollina-tors (Shirley 1996)Over half of polyploids have an unexpected inheritance pat-
tern for chlorophyll if it is assumed that the presence of chloro-phyll is a dominant character Those polyploids that deviatefrom expectation span an age range of synthetic to 10myoand always lack chlorophyll in their petals which is in linewith the direction of the shifts observed in the divergence ofdiploid species (Supplementary Data Fig S6) In carnation thedifference in chlorophyll concentration between white andgreen flowers is likely caused by downregulation of chlorophyllbiosynthesis genes in white flowers genes involved in
TABLE 1 Polyploid and homoploid hybrid observed and expected floral colours
Species Spectral Bee Hummingbird Chlorophyll
Observed Expected Observed Expected Observed Expected Observed Expected
N tabacum 095-55 R W P UV-B B-G sUV-P W P N CN tabacum 51789 P W P B-G B-G P W P N CN tabacum lsquoChulumanirsquo W W P B-G B-G P W P N Csyn N tabacum QM P W P B B-G LP W P N Csyn N tabacum TH37 P W P B-G B-G P W P N CTH32 P W P B-G B-G P W P N CN rustica var asiatica G G LG G B-G G G W C CN rustica var pavonii G G G G B-G G G W C Csyn UP (homoploid) G G UV-G G B-G UV-G G W C Csyn F1 (homoploid) G G LG G B-G G G W C Csyn N rustica S0 G G LG G B-G G G W C Csyn N rustica S1 G G LG G B-G G G W C CN arentsii W G W B-G B-G P W C CN clevelandii W UV-W W B-G UV B-G P UV-W W N CN quadrivalvis TW18 W UV-W W B-G UV B-G W UV-W W C CN quadrivalvis 9047 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 1 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 2 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 5 W UV-W W B-G UV B-G W UV-W W C CN repanda W W UV-W B-G B-G UV W W UV-W N CN nesophila W W UV-W B-G B-G UV W W UV-W N CN stocktonii W W UV-W B-G B-G UV W W UV-W C CN nudicaulis UV-W W UV-W UV B-G UV UV-W W UV-W C CN benthamiana W W ndash ndash ndash ndash N CN forsteri W W ndash ndash ndash ndash C CN gossei W W ndash ndash ndash ndash N CN megalosiphon W W ndash ndash ndash ndash N CN occidentalis W W ndash ndash ndash ndash N CN suaveolens W W ndash ndash ndash ndash N C
N glauca 51725 Y W ndash ndash ndash ndash C CN glauca 51751 YG W ndash ndash ndash ndash C CN linearis TW77 UV-W W ndash ndash ndash ndash C CN linearis 9647 UV-W W ndash ndash ndash ndash C CN glutinosa P P W G ndash ndash ndash ndash C C
The top block consists of polyploid accessions and the bottom block includes homoploid hybridsItalic denotes an unexpected phenotype given the colour categories of the progenitorsProgenitor bee and hummingbird colour categories are unknown for section Suaveolentes and natural homoploid hybrids (see text)R red W white P pink G green UV-W UVndashwhite Y yellow UV-B UVndashblue B-G bluendashgreen B blue LG light green UV high UV UV-G
UVndashgreen sUV-P saturated UVndashpink LP light pink N no chlorophyll C chlorophyll syn synthetic 9047 represents 904750042 9647 represents964750099
1126 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
resulted in 11 colour categories which fell into the followingareas of bee colour space saturated green UVndashblue high UVUVndashgreen green light green bluendashgreen dark green saturatedUVndashblue saturated UVndashgreen and blue (the last four categorieswere each represented by only a single accessionSupplementary Data Fig S3A) These groups are shown in thebee colour hexagon (Fig 4C) The hummingbird colour spacecluster analysis also produced 11 colour categories saturatedgreen green UVndashwhite UVndashgreen pink white UVndashpinkdark green light pink red and saturated UVndashpink (again thelast four categories include only a single accessionSupplementary Data Fig S3B) These groups are shown in thehummingbird colour space (Fig 4D) and the same graph isprovided as an animation to better display the 3D nature of thecolour space (Supplementary Data Video)
Evolution of spectral reflectance in polyploids and homoploidhybrids
To assess the evolution of polyploid floral colour polyploidand homoploid hybrid accessions were compared with those oftheir progenitors in spectral bee and hummingbird colour cate-gories as well as in the presenceabsence of chlorophyll Thediploid progenitors and approximate age of polyploids andhomoploid hybrids are found in Fig 1 and Supplementary DataTable S4 and the observed and expected floral colours of
polyploid and homoploid hybrids are found in Table 1 Mostpolyploid and homoploid hybrids were similar to at least oneprogenitor in spectral bee and hummingbird colour categoriesbut some fell into unexpected colour categories (Table 1Fig 5 Supplementary Data Fig S4 Supplementary Data FigS5) Over half of the polyploids unexpectedly lacked chloro-phyll (Table 1)
Evolution of colour characters in a phylogenetic context
Reconstructed character states are shown for spectral reflec-tance colour categories (Fig 6) and the presenceabsence ofchlorophyll in petals (Supplementary Data Fig S6) Bee andhummingbird colour categories are also shown for extant spe-cies on the plastid tree (Fig 6) Although the deepest nodeswere largely equivocal evolution of spectral reflectance colourin Nicotiana seemed to be dynamic (Fig 6) Green flowerslikely have three independent origins (1) in sectionsPaniculatae and Undulatae (2) in N langsdorffii and (3) inthe homoploid hybrid N glauca UVndashwhite flowers also seemto have arisen three times independently (1) in sectionTrigonophyllae (2) in N pauciflora and (3) in the homoploidhybrid N linearis Most polyploid and homoploid hybrid spe-cies exhibit a floral colour present in at least one of their pro-genitors However N tabacum 095-55 is red and N glauca isyellow and green unlike their progenitors UVndashwhite flowersseem to have evolved de novo in N linearis UVndashwhite flowersare also found in one of its progenitor sections (in N pauci-flora) but ancestral reconstructions indicate that the floral col-our was most likely white at the ancestral nodes within thesection (Fig 6) This suggests that the evolution of UVndashwhiteflowers in N pauciflora has occurred subsequent to the forma-tion of N linearis and that the two events are likely indepen-dent It is unclear whether UVndashwhite flowers also evolved denovo in N nudicaulis because the ancestral node of sectionRepandae is equivocal The presence of chlorophyll as inferredby light absorption at 675 nm (Haardt and Maske 1987) inNicotiana flowers is ancestral and has been lost three times inN sylvestris N noctiflora and the most recent common ances-tor of N acuminata and N pauciflora (Supplementary DataFig S6)Results from Mantel tests of phylogenetic signal for
Nicotiana floral traits for both genetic distance and the 36 000post-burn-in Bayesian trees are shown in Table 2 All floraltraits were significantly correlated with phylogenetic relation-ships for the Bayesian trees at a significance level of Plt 005Only spectral reflectance was significant for the genetic dis-tance tests whereas bee and hummingbird colour perceptionwere just above the Plt 005 threshold For the Bayesian trees901 662 and 932 of trees were significantly correlatedwith the spectral reflectance bee and hummingbird colour per-ception datasets respectively These results suggest that thesefloral traits are weakly constrained by phylogeny
DISCUSSION
Nicotiana is remarkable in its range of spectral reflectanceflower colours (white UVndashwhite pink magenta red yellowgreen and dark green Fig 3) and in the variety of pollinators
0
200
400
600
800
1000
1200C
ell
are
a (
microm
2)
N rustica
N paniculata
N undulata
N sylvestris
N obtusifolia
N nudicaulis
N repanda
N stocktonii
N rustica var asiatica
N rustica var pavonii
Repandae
a
a
b
b
cc
c c
d
Maternal progenitor
Paternal progenitor
Polyploids
Polyploids
Polyploids
FIG 2 Petal cell area from polyploids and their progenitors Within each poly-ploid group bars with different letters represent significantly different mean cell
areas
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1123
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
that visit the flowers (moth bird bee bat Knapp 2010) Theperception of these spectral colours also changes with visualsystem (bee or hummingbird) Here we describe a complex dy-namic in the evolution of floral colour in Nicotiana Spectral re-flectance and bee and hummingbird colour perception arecorrelated with phylogeny but multiple independent origins ofvarious combinations of pigmentation suggest that the evolutionof floral colour is not entirely phylogenetically constrained
Petal cell size evolution in polyploids
Cell size is expected to increase following polyploidizationdue to the increase in genome size (Beaulieu et al 2008) Thesignificantly larger petal cells of N rustica (lt02 million yearsold myo) and the intermediate cell size of section Repandae(45myo) suggest that polyploids may revert to a diploid-like cell size over time similar to the genome downsizing ob-served in polyploids (Leitch and Bennett 2004) Howeverpetal cell size differences within section Repandae do not seemto be linked to genome size N nudicaulis and N repanda sharesimilar cell sizes but have substantially different genome sizes(Leitch et al 2008) The concentration of pigment in floralcells is controlled by both the amount of pigment present andthe cell size Nicotiana rustica has significantly larger petalcells than its progenitors (close to the sum Fig 2) and displaysan intermediate brightness (the area under the reflectance curvea proxy for pigment concentration) between its progenitors
(Supplementary Data Fig S4D) This is expected if the poly-ploid inherits the sum of both cell size and amount of pigmentpresent from its progenitors
Polyploid divergence in floral colour
Many younger polyploids (lt02myo) display unexpectedfloral colours considering those of their diploid progenitorsNone of the natural and synthetic N tabacum accessions pos-sess chlorophyll which is unexpected given its presence in atleast one progenitor species Nicotiana tabacum 095-55 alsohas unexpected spectral bee and hummingbird colour giventhe colour categories of the progenitor species Similarly syn-thetic N tabacum QM has unexpected colours in bee and hum-mingbird perception Because this accession is synthetic theparents are known and thus its unexpected phenotype can beclassified as transgressive or outside the range of its progeni-tors due solely to polyploidy and hybridization Most N rusticaaccessions have unexpected bee colour (four of these aresynthetic and are therefore also transgressive) and N arentsiihas unexpected hummingbird colour (Table 1 Fig 6) Despitethe divergence of floral spectra associated with polyploidybehavioural studies are needed to determine whether the beeand hummingbird colour categories delineated here actuallyelicit different responses in pollinatorsMost older polyploids (1ndash10myo) are similar in floral col-
our category to at least one of their progenitors N clevelandii
synth
etic
N ru
stic
a P
UE
1-R
1 S
1
N ru
stic
a v
ar a
sia
tica
synth
etic
N ru
stic
a P
UE
1-R
10 S
0
synth
etic
PU
E1 F
1synth
etic
UxP
N b
enavid
esii
N ru
stic
a v
ar p
avonii
N p
anic
ula
taN
langsdorffii
N g
lauca 5
1751 g
reen
N u
ndula
taN
raim
ondii
N k
nig
htia
na
N g
lauca 5
1751 y
ello
w
N g
lauca 5
1725
N p
lum
bagin
ifolia
N g
ossei
N q
uadriv
alv
is T
W18
N o
ccid
enta
lis s
ubsp h
esperis
N
megalo
sip
hon
N s
uaveole
ns
N n
esophila
N s
tockto
nii
N q
uadriv
alv
is 9
04750042
N a
ttenuata
N re
panda
N s
ylv
estris
A04750326
N
benth
am
iana
N m
uta
bilis
CP
G3 w
hite
N
muta
bilis
CP
G12456 w
hite
N
wig
andio
ides
N a
cum
inata
N fo
rste
riN
cle
vela
ndii
N n
octiflo
raN
x o
btu
sia
ta lin
e 2
N
x o
btu
sia
ta lin
e 1
N
mie
rsii
N p
etu
nio
ides
N x
obtu
sia
ta lin
e 5
N
sylv
estris
6898
N
oto
phora
white
N
are
nts
iiT
H3
2N
tabacum
lsquoChulu
manirsquo
N o
btu
sifo
lia v
ar p
alm
eri
N o
btu
sifo
lia v
ar o
btu
sifo
lia lsquoB
ald
win
rsquoN
linearis
TW
77
N n
udic
aulis
N o
btu
sifo
lia v
ar o
btu
sifo
lia T
W143
N lin
earis
964750099
N
pauciflo
raN
tom
ento
sifo
rmis
N o
tophora
pin
k
N g
lutin
osa
synth
etic
N ta
bacum
TH
37
N
tabacum
51789
synth
etic
N ta
bacum
QM
N
tabacum
095-5
5
N s
etc
hellii
N m
uta
bilis
CP
G3 p
ink
N m
uta
bilis
CP
G12456 p
ink
10
8
6
4
2
0
Dis
tan
ce
Spectral categories
MagentaRedPinkUV-white
WhiteYellowDark greenGreen
Dark greenSaturated greenGreenUV-whiteUV-green
PinkLight pinkWhiteUV-pinkRedSaturated UV-pink
SpecBeeHum
SpectralDark greenSaturated greenSaturated UV-blueUV-blueSaturated UV-greenHigh UV
UV-greenGreenLight greenBlueBlue-green
Bee Hummingbird
FIG 3 Dendrograms based on distance cluster analyses for spectral reflectance Coloured circles on the dendrogram represent distinct colour categories as determinedby the chosen threshold (dashed line) Dendrograms were similarly constructed for bee and hummingbird colour (see Supplementary Data Fig S3) The lines of col-oured circles at the tips of the dendrograms signify the category each taxon is assigned to in spectral bee and hummingbird colour as labelled (Spec Bee and Hum)for comparison between spectral categories and those of different visual systems Diploids polyploids and homoploids are denoted by black blue and orange text
respectively
1124 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
0
0middot2
0middot4
0middot6
0middot8
1middot0
Reflecta
nce
Wavelength (nm)
0
0middot2
0middot4
0middot6
0middot8
1middot0
300 350 400 450 500 550 600 650 700 300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
N benavidesii
N glauca 51751 g
N langsdorffii
N paniculata
N raimondii
N rustica var asiatica
N rustica var pavonii
synthetic UxP
synthetic F1
syn N rustica S0
syn N rustica S1
N undulata
N glutinosa
N otophora p
N tabacum 51789
syn N tabacum QM
syn N tabacum TH37
N tomentosiformis
A B
G 50
R 50
B 50
UV 50
Hummingbird colour space
White
0middot40middot4
ndash0middot40middot4
0middot4
0middot4
y
z
UV-white
Pink
Synthetic N tabacum QM
UV-pink
N tabacum 095-55
N setchellii
Green
N langsdorffii
Saturated green
UV-green
C DBee colour hexagon
Blue-green
Light green
Green
Dark green
Sat green
UV-green
Sat UV-green
B
B-G
G
UV-G
UV
UV-B
1 acum2 aren3 atten4 bena5 benth6 clev7 forst8 glau259 glau51y10 glau51g11 glut12 goss13 knight14 langs15 lin964716 linTW7717 mega18 mier19 mut1w20 mut1p21 mut3w22mut3p23 neso24 noct25 nudi26 xobtus127 xobtus228 xobtus529 obtusB30 obtusTW31 obtuspalm
32 occhesp33 otoph w34 otoph p35 pani36 pauc37 petun38 plumba39 quad904740 quadTW1841 raim42 repa43 rustasi44 rustpav45 syn UxP46 syn F147 syn rust S048 syn rust S149 setch50 stock51 suav52 sylv689853 sylvA04754 tab0955555 tab5178956 tabchulu57 syntabQM58 syntabTH3759 tomtform60 undu61 wigan62TH32
High UV
UV-blue
Sat UV-blue
Blue
FIG 4 (A B) Nicotiana reflectance spectra from 300 to 700nm which roughly correspond to colours perceived by human observers as pink (A) and green (B) SeeSupplementary Data Fig S2 for other spectral colour categories Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploidhybrid taxa Abbreviations p pink syn synthetic g green (C) Colour hexagon displaying the distribution of Nicotiana colour loci in bee colour space The hexa-gon has been scaled so that vertices represent 40 excitation of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-GUVndashgreen Bee colour categories are delineated by coloured ovals sat saturated Nicotiana species abbreviations are as follows acum acuminata aren arentsiiatten attenuata benavid benavidesii benth benthamiana clev clevelandii forst forsteri glau25 glauca 51725 glau51y glauca 51751 yellow glau51g glauca51751 green glut glutinosa goss gossei knight knightiana langs langsdorffii lin9647 linearis 964750099 linTW77 linearis TW77 mega megalosiphon miermiersii mut1w mutabilis CPG12456 white mut1p mutabilis CPG12456 pink mut3w mutabilis CPG3 white mut3p mutabilis CPG3 pink neso nesophila noctnoctiflora nudi nudicaulis obtus1 obtusiata line 1 obtus2 obtusiata line 2 obtus5 obtusiata line 5 obtusB obtusifolia var obtusifolia lsquoBaldwinrsquoobtusTW obtusifolia var obtusifolia TW143 obtuspalm obtusifolia var palmeri occhesp occidentalis subsp hesperis otoph w otophora white otoph p otophorapink pani paniculata pauc pauciflora petun petunioides plumba plumbaginifolia quad9047 quadrivalvis 904750042 quadTW18 quadrivalvis TW18 raimraimondii repa repanda rustasi rustica var asiatica rustpav rustica var pavonii syn UP synthetic UP syn F1 synthetic PUE1 F1 synrust S0 synthetic rus-tica PUE1-R10 S0 synrust S1 synthetic rustica PUE1-R1 S1 setch setchellii stock stocktonii suav suaveolens sylv6898 sylvestris 6898 sylvA047 sylvestrisA04750326 tab09555 tabacum 095-55 tab51789 tabacum 51789 tabchulu tabacum lsquoChulumanirsquo syntabQM synthetic tabacum QM syntabTH37 synthetictabacum TH37 tomtform tomentosiformis undu undulata wigan wigandioides TH32 TH32 synthetic N sylvestrisN otophora polyploid (D) Distribution ofNicotiana spectral loci in hummingbird colour space Vertices of the hummingbird colour space represent 50 excitation of the photoreceptors single photoreceptortype vertices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows repre-sent the vectors of these photoreceptors from the origin of the hummingbird colour space Nicotiana loci are coloured according to hummingbird colour categories(Supplementary Data Fig S3B) but are labelled with the accession name if the category includes only one taxon See Supplementary Data Video for an animation
of this graph
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1125
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
is the exception because it falls into an unexpected humming-bird colour category given the progenitor species (Fig 6)Section Repandae polyploids seem to have evolved to be eitherlike their maternal (Nicotiana nesophila N repanda andN stocktonii) or paternal (N nudicaulis) progenitor (Fig 6)The maternal progenitor N sylvestris is no longer sympatricwith any of the section Repandae polyploids therefore Nnesophila N repanda and N stocktonii can occupy the samepollination niche as their maternal progenitor without competi-tion Similarly section Suaveolentes is native to Australasiaexcept for one species in Namibia Africa and is not sympatricwith its progenitor sections in South America (Goodspeed1954) these polyploids and their diploid progenitors displaysimilar floral colours except N pauciflora which evolvedspectrally UVndashwhite flowers after the formation of sectionSuaveolentes (Fig 6) It is possible that there is less competitionfor pollinators and therefore reduced selective pressure to-wards floral colour diversification when polyploid species arenot sympatric with their diploid progenitors as is seen in
Iochrominae (Solanaceae) which have a broader range of floralcolours when species are sympatric (Muchhala et al 2014)However floral colour evolution can also be driven by geneticdrift or selection on pleiotropic effects of floral genes (Chittkaet al 2001 Rausher 2008) Furthermore anthocyanins andflavonoids are important for UV protection can defend againstfungi act as signal molecules and play a role in male fertility insome species in addition to their roles in signalling to pollina-tors (Shirley 1996)Over half of polyploids have an unexpected inheritance pat-
tern for chlorophyll if it is assumed that the presence of chloro-phyll is a dominant character Those polyploids that deviatefrom expectation span an age range of synthetic to 10myoand always lack chlorophyll in their petals which is in linewith the direction of the shifts observed in the divergence ofdiploid species (Supplementary Data Fig S6) In carnation thedifference in chlorophyll concentration between white andgreen flowers is likely caused by downregulation of chlorophyllbiosynthesis genes in white flowers genes involved in
TABLE 1 Polyploid and homoploid hybrid observed and expected floral colours
Species Spectral Bee Hummingbird Chlorophyll
Observed Expected Observed Expected Observed Expected Observed Expected
N tabacum 095-55 R W P UV-B B-G sUV-P W P N CN tabacum 51789 P W P B-G B-G P W P N CN tabacum lsquoChulumanirsquo W W P B-G B-G P W P N Csyn N tabacum QM P W P B B-G LP W P N Csyn N tabacum TH37 P W P B-G B-G P W P N CTH32 P W P B-G B-G P W P N CN rustica var asiatica G G LG G B-G G G W C CN rustica var pavonii G G G G B-G G G W C Csyn UP (homoploid) G G UV-G G B-G UV-G G W C Csyn F1 (homoploid) G G LG G B-G G G W C Csyn N rustica S0 G G LG G B-G G G W C Csyn N rustica S1 G G LG G B-G G G W C CN arentsii W G W B-G B-G P W C CN clevelandii W UV-W W B-G UV B-G P UV-W W N CN quadrivalvis TW18 W UV-W W B-G UV B-G W UV-W W C CN quadrivalvis 9047 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 1 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 2 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 5 W UV-W W B-G UV B-G W UV-W W C CN repanda W W UV-W B-G B-G UV W W UV-W N CN nesophila W W UV-W B-G B-G UV W W UV-W N CN stocktonii W W UV-W B-G B-G UV W W UV-W C CN nudicaulis UV-W W UV-W UV B-G UV UV-W W UV-W C CN benthamiana W W ndash ndash ndash ndash N CN forsteri W W ndash ndash ndash ndash C CN gossei W W ndash ndash ndash ndash N CN megalosiphon W W ndash ndash ndash ndash N CN occidentalis W W ndash ndash ndash ndash N CN suaveolens W W ndash ndash ndash ndash N C
N glauca 51725 Y W ndash ndash ndash ndash C CN glauca 51751 YG W ndash ndash ndash ndash C CN linearis TW77 UV-W W ndash ndash ndash ndash C CN linearis 9647 UV-W W ndash ndash ndash ndash C CN glutinosa P P W G ndash ndash ndash ndash C C
The top block consists of polyploid accessions and the bottom block includes homoploid hybridsItalic denotes an unexpected phenotype given the colour categories of the progenitorsProgenitor bee and hummingbird colour categories are unknown for section Suaveolentes and natural homoploid hybrids (see text)R red W white P pink G green UV-W UVndashwhite Y yellow UV-B UVndashblue B-G bluendashgreen B blue LG light green UV high UV UV-G
UVndashgreen sUV-P saturated UVndashpink LP light pink N no chlorophyll C chlorophyll syn synthetic 9047 represents 904750042 9647 represents964750099
1126 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
that visit the flowers (moth bird bee bat Knapp 2010) Theperception of these spectral colours also changes with visualsystem (bee or hummingbird) Here we describe a complex dy-namic in the evolution of floral colour in Nicotiana Spectral re-flectance and bee and hummingbird colour perception arecorrelated with phylogeny but multiple independent origins ofvarious combinations of pigmentation suggest that the evolutionof floral colour is not entirely phylogenetically constrained
Petal cell size evolution in polyploids
Cell size is expected to increase following polyploidizationdue to the increase in genome size (Beaulieu et al 2008) Thesignificantly larger petal cells of N rustica (lt02 million yearsold myo) and the intermediate cell size of section Repandae(45myo) suggest that polyploids may revert to a diploid-like cell size over time similar to the genome downsizing ob-served in polyploids (Leitch and Bennett 2004) Howeverpetal cell size differences within section Repandae do not seemto be linked to genome size N nudicaulis and N repanda sharesimilar cell sizes but have substantially different genome sizes(Leitch et al 2008) The concentration of pigment in floralcells is controlled by both the amount of pigment present andthe cell size Nicotiana rustica has significantly larger petalcells than its progenitors (close to the sum Fig 2) and displaysan intermediate brightness (the area under the reflectance curvea proxy for pigment concentration) between its progenitors
(Supplementary Data Fig S4D) This is expected if the poly-ploid inherits the sum of both cell size and amount of pigmentpresent from its progenitors
Polyploid divergence in floral colour
Many younger polyploids (lt02myo) display unexpectedfloral colours considering those of their diploid progenitorsNone of the natural and synthetic N tabacum accessions pos-sess chlorophyll which is unexpected given its presence in atleast one progenitor species Nicotiana tabacum 095-55 alsohas unexpected spectral bee and hummingbird colour giventhe colour categories of the progenitor species Similarly syn-thetic N tabacum QM has unexpected colours in bee and hum-mingbird perception Because this accession is synthetic theparents are known and thus its unexpected phenotype can beclassified as transgressive or outside the range of its progeni-tors due solely to polyploidy and hybridization Most N rusticaaccessions have unexpected bee colour (four of these aresynthetic and are therefore also transgressive) and N arentsiihas unexpected hummingbird colour (Table 1 Fig 6) Despitethe divergence of floral spectra associated with polyploidybehavioural studies are needed to determine whether the beeand hummingbird colour categories delineated here actuallyelicit different responses in pollinatorsMost older polyploids (1ndash10myo) are similar in floral col-
our category to at least one of their progenitors N clevelandii
synth
etic
N ru
stic
a P
UE
1-R
1 S
1
N ru
stic
a v
ar a
sia
tica
synth
etic
N ru
stic
a P
UE
1-R
10 S
0
synth
etic
PU
E1 F
1synth
etic
UxP
N b
enavid
esii
N ru
stic
a v
ar p
avonii
N p
anic
ula
taN
langsdorffii
N g
lauca 5
1751 g
reen
N u
ndula
taN
raim
ondii
N k
nig
htia
na
N g
lauca 5
1751 y
ello
w
N g
lauca 5
1725
N p
lum
bagin
ifolia
N g
ossei
N q
uadriv
alv
is T
W18
N o
ccid
enta
lis s
ubsp h
esperis
N
megalo
sip
hon
N s
uaveole
ns
N n
esophila
N s
tockto
nii
N q
uadriv
alv
is 9
04750042
N a
ttenuata
N re
panda
N s
ylv
estris
A04750326
N
benth
am
iana
N m
uta
bilis
CP
G3 w
hite
N
muta
bilis
CP
G12456 w
hite
N
wig
andio
ides
N a
cum
inata
N fo
rste
riN
cle
vela
ndii
N n
octiflo
raN
x o
btu
sia
ta lin
e 2
N
x o
btu
sia
ta lin
e 1
N
mie
rsii
N p
etu
nio
ides
N x
obtu
sia
ta lin
e 5
N
sylv
estris
6898
N
oto
phora
white
N
are
nts
iiT
H3
2N
tabacum
lsquoChulu
manirsquo
N o
btu
sifo
lia v
ar p
alm
eri
N o
btu
sifo
lia v
ar o
btu
sifo
lia lsquoB
ald
win
rsquoN
linearis
TW
77
N n
udic
aulis
N o
btu
sifo
lia v
ar o
btu
sifo
lia T
W143
N lin
earis
964750099
N
pauciflo
raN
tom
ento
sifo
rmis
N o
tophora
pin
k
N g
lutin
osa
synth
etic
N ta
bacum
TH
37
N
tabacum
51789
synth
etic
N ta
bacum
QM
N
tabacum
095-5
5
N s
etc
hellii
N m
uta
bilis
CP
G3 p
ink
N m
uta
bilis
CP
G12456 p
ink
10
8
6
4
2
0
Dis
tan
ce
Spectral categories
MagentaRedPinkUV-white
WhiteYellowDark greenGreen
Dark greenSaturated greenGreenUV-whiteUV-green
PinkLight pinkWhiteUV-pinkRedSaturated UV-pink
SpecBeeHum
SpectralDark greenSaturated greenSaturated UV-blueUV-blueSaturated UV-greenHigh UV
UV-greenGreenLight greenBlueBlue-green
Bee Hummingbird
FIG 3 Dendrograms based on distance cluster analyses for spectral reflectance Coloured circles on the dendrogram represent distinct colour categories as determinedby the chosen threshold (dashed line) Dendrograms were similarly constructed for bee and hummingbird colour (see Supplementary Data Fig S3) The lines of col-oured circles at the tips of the dendrograms signify the category each taxon is assigned to in spectral bee and hummingbird colour as labelled (Spec Bee and Hum)for comparison between spectral categories and those of different visual systems Diploids polyploids and homoploids are denoted by black blue and orange text
respectively
1124 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
0
0middot2
0middot4
0middot6
0middot8
1middot0
Reflecta
nce
Wavelength (nm)
0
0middot2
0middot4
0middot6
0middot8
1middot0
300 350 400 450 500 550 600 650 700 300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
N benavidesii
N glauca 51751 g
N langsdorffii
N paniculata
N raimondii
N rustica var asiatica
N rustica var pavonii
synthetic UxP
synthetic F1
syn N rustica S0
syn N rustica S1
N undulata
N glutinosa
N otophora p
N tabacum 51789
syn N tabacum QM
syn N tabacum TH37
N tomentosiformis
A B
G 50
R 50
B 50
UV 50
Hummingbird colour space
White
0middot40middot4
ndash0middot40middot4
0middot4
0middot4
y
z
UV-white
Pink
Synthetic N tabacum QM
UV-pink
N tabacum 095-55
N setchellii
Green
N langsdorffii
Saturated green
UV-green
C DBee colour hexagon
Blue-green
Light green
Green
Dark green
Sat green
UV-green
Sat UV-green
B
B-G
G
UV-G
UV
UV-B
1 acum2 aren3 atten4 bena5 benth6 clev7 forst8 glau259 glau51y10 glau51g11 glut12 goss13 knight14 langs15 lin964716 linTW7717 mega18 mier19 mut1w20 mut1p21 mut3w22mut3p23 neso24 noct25 nudi26 xobtus127 xobtus228 xobtus529 obtusB30 obtusTW31 obtuspalm
32 occhesp33 otoph w34 otoph p35 pani36 pauc37 petun38 plumba39 quad904740 quadTW1841 raim42 repa43 rustasi44 rustpav45 syn UxP46 syn F147 syn rust S048 syn rust S149 setch50 stock51 suav52 sylv689853 sylvA04754 tab0955555 tab5178956 tabchulu57 syntabQM58 syntabTH3759 tomtform60 undu61 wigan62TH32
High UV
UV-blue
Sat UV-blue
Blue
FIG 4 (A B) Nicotiana reflectance spectra from 300 to 700nm which roughly correspond to colours perceived by human observers as pink (A) and green (B) SeeSupplementary Data Fig S2 for other spectral colour categories Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploidhybrid taxa Abbreviations p pink syn synthetic g green (C) Colour hexagon displaying the distribution of Nicotiana colour loci in bee colour space The hexa-gon has been scaled so that vertices represent 40 excitation of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-GUVndashgreen Bee colour categories are delineated by coloured ovals sat saturated Nicotiana species abbreviations are as follows acum acuminata aren arentsiiatten attenuata benavid benavidesii benth benthamiana clev clevelandii forst forsteri glau25 glauca 51725 glau51y glauca 51751 yellow glau51g glauca51751 green glut glutinosa goss gossei knight knightiana langs langsdorffii lin9647 linearis 964750099 linTW77 linearis TW77 mega megalosiphon miermiersii mut1w mutabilis CPG12456 white mut1p mutabilis CPG12456 pink mut3w mutabilis CPG3 white mut3p mutabilis CPG3 pink neso nesophila noctnoctiflora nudi nudicaulis obtus1 obtusiata line 1 obtus2 obtusiata line 2 obtus5 obtusiata line 5 obtusB obtusifolia var obtusifolia lsquoBaldwinrsquoobtusTW obtusifolia var obtusifolia TW143 obtuspalm obtusifolia var palmeri occhesp occidentalis subsp hesperis otoph w otophora white otoph p otophorapink pani paniculata pauc pauciflora petun petunioides plumba plumbaginifolia quad9047 quadrivalvis 904750042 quadTW18 quadrivalvis TW18 raimraimondii repa repanda rustasi rustica var asiatica rustpav rustica var pavonii syn UP synthetic UP syn F1 synthetic PUE1 F1 synrust S0 synthetic rus-tica PUE1-R10 S0 synrust S1 synthetic rustica PUE1-R1 S1 setch setchellii stock stocktonii suav suaveolens sylv6898 sylvestris 6898 sylvA047 sylvestrisA04750326 tab09555 tabacum 095-55 tab51789 tabacum 51789 tabchulu tabacum lsquoChulumanirsquo syntabQM synthetic tabacum QM syntabTH37 synthetictabacum TH37 tomtform tomentosiformis undu undulata wigan wigandioides TH32 TH32 synthetic N sylvestrisN otophora polyploid (D) Distribution ofNicotiana spectral loci in hummingbird colour space Vertices of the hummingbird colour space represent 50 excitation of the photoreceptors single photoreceptortype vertices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows repre-sent the vectors of these photoreceptors from the origin of the hummingbird colour space Nicotiana loci are coloured according to hummingbird colour categories(Supplementary Data Fig S3B) but are labelled with the accession name if the category includes only one taxon See Supplementary Data Video for an animation
of this graph
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1125
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
is the exception because it falls into an unexpected humming-bird colour category given the progenitor species (Fig 6)Section Repandae polyploids seem to have evolved to be eitherlike their maternal (Nicotiana nesophila N repanda andN stocktonii) or paternal (N nudicaulis) progenitor (Fig 6)The maternal progenitor N sylvestris is no longer sympatricwith any of the section Repandae polyploids therefore Nnesophila N repanda and N stocktonii can occupy the samepollination niche as their maternal progenitor without competi-tion Similarly section Suaveolentes is native to Australasiaexcept for one species in Namibia Africa and is not sympatricwith its progenitor sections in South America (Goodspeed1954) these polyploids and their diploid progenitors displaysimilar floral colours except N pauciflora which evolvedspectrally UVndashwhite flowers after the formation of sectionSuaveolentes (Fig 6) It is possible that there is less competitionfor pollinators and therefore reduced selective pressure to-wards floral colour diversification when polyploid species arenot sympatric with their diploid progenitors as is seen in
Iochrominae (Solanaceae) which have a broader range of floralcolours when species are sympatric (Muchhala et al 2014)However floral colour evolution can also be driven by geneticdrift or selection on pleiotropic effects of floral genes (Chittkaet al 2001 Rausher 2008) Furthermore anthocyanins andflavonoids are important for UV protection can defend againstfungi act as signal molecules and play a role in male fertility insome species in addition to their roles in signalling to pollina-tors (Shirley 1996)Over half of polyploids have an unexpected inheritance pat-
tern for chlorophyll if it is assumed that the presence of chloro-phyll is a dominant character Those polyploids that deviatefrom expectation span an age range of synthetic to 10myoand always lack chlorophyll in their petals which is in linewith the direction of the shifts observed in the divergence ofdiploid species (Supplementary Data Fig S6) In carnation thedifference in chlorophyll concentration between white andgreen flowers is likely caused by downregulation of chlorophyllbiosynthesis genes in white flowers genes involved in
TABLE 1 Polyploid and homoploid hybrid observed and expected floral colours
Species Spectral Bee Hummingbird Chlorophyll
Observed Expected Observed Expected Observed Expected Observed Expected
N tabacum 095-55 R W P UV-B B-G sUV-P W P N CN tabacum 51789 P W P B-G B-G P W P N CN tabacum lsquoChulumanirsquo W W P B-G B-G P W P N Csyn N tabacum QM P W P B B-G LP W P N Csyn N tabacum TH37 P W P B-G B-G P W P N CTH32 P W P B-G B-G P W P N CN rustica var asiatica G G LG G B-G G G W C CN rustica var pavonii G G G G B-G G G W C Csyn UP (homoploid) G G UV-G G B-G UV-G G W C Csyn F1 (homoploid) G G LG G B-G G G W C Csyn N rustica S0 G G LG G B-G G G W C Csyn N rustica S1 G G LG G B-G G G W C CN arentsii W G W B-G B-G P W C CN clevelandii W UV-W W B-G UV B-G P UV-W W N CN quadrivalvis TW18 W UV-W W B-G UV B-G W UV-W W C CN quadrivalvis 9047 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 1 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 2 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 5 W UV-W W B-G UV B-G W UV-W W C CN repanda W W UV-W B-G B-G UV W W UV-W N CN nesophila W W UV-W B-G B-G UV W W UV-W N CN stocktonii W W UV-W B-G B-G UV W W UV-W C CN nudicaulis UV-W W UV-W UV B-G UV UV-W W UV-W C CN benthamiana W W ndash ndash ndash ndash N CN forsteri W W ndash ndash ndash ndash C CN gossei W W ndash ndash ndash ndash N CN megalosiphon W W ndash ndash ndash ndash N CN occidentalis W W ndash ndash ndash ndash N CN suaveolens W W ndash ndash ndash ndash N C
N glauca 51725 Y W ndash ndash ndash ndash C CN glauca 51751 YG W ndash ndash ndash ndash C CN linearis TW77 UV-W W ndash ndash ndash ndash C CN linearis 9647 UV-W W ndash ndash ndash ndash C CN glutinosa P P W G ndash ndash ndash ndash C C
The top block consists of polyploid accessions and the bottom block includes homoploid hybridsItalic denotes an unexpected phenotype given the colour categories of the progenitorsProgenitor bee and hummingbird colour categories are unknown for section Suaveolentes and natural homoploid hybrids (see text)R red W white P pink G green UV-W UVndashwhite Y yellow UV-B UVndashblue B-G bluendashgreen B blue LG light green UV high UV UV-G
UVndashgreen sUV-P saturated UVndashpink LP light pink N no chlorophyll C chlorophyll syn synthetic 9047 represents 904750042 9647 represents964750099
1126 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
0
0middot2
0middot4
0middot6
0middot8
1middot0
Reflecta
nce
Wavelength (nm)
0
0middot2
0middot4
0middot6
0middot8
1middot0
300 350 400 450 500 550 600 650 700 300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
N benavidesii
N glauca 51751 g
N langsdorffii
N paniculata
N raimondii
N rustica var asiatica
N rustica var pavonii
synthetic UxP
synthetic F1
syn N rustica S0
syn N rustica S1
N undulata
N glutinosa
N otophora p
N tabacum 51789
syn N tabacum QM
syn N tabacum TH37
N tomentosiformis
A B
G 50
R 50
B 50
UV 50
Hummingbird colour space
White
0middot40middot4
ndash0middot40middot4
0middot4
0middot4
y
z
UV-white
Pink
Synthetic N tabacum QM
UV-pink
N tabacum 095-55
N setchellii
Green
N langsdorffii
Saturated green
UV-green
C DBee colour hexagon
Blue-green
Light green
Green
Dark green
Sat green
UV-green
Sat UV-green
B
B-G
G
UV-G
UV
UV-B
1 acum2 aren3 atten4 bena5 benth6 clev7 forst8 glau259 glau51y10 glau51g11 glut12 goss13 knight14 langs15 lin964716 linTW7717 mega18 mier19 mut1w20 mut1p21 mut3w22mut3p23 neso24 noct25 nudi26 xobtus127 xobtus228 xobtus529 obtusB30 obtusTW31 obtuspalm
32 occhesp33 otoph w34 otoph p35 pani36 pauc37 petun38 plumba39 quad904740 quadTW1841 raim42 repa43 rustasi44 rustpav45 syn UxP46 syn F147 syn rust S048 syn rust S149 setch50 stock51 suav52 sylv689853 sylvA04754 tab0955555 tab5178956 tabchulu57 syntabQM58 syntabTH3759 tomtform60 undu61 wigan62TH32
High UV
UV-blue
Sat UV-blue
Blue
FIG 4 (A B) Nicotiana reflectance spectra from 300 to 700nm which roughly correspond to colours perceived by human observers as pink (A) and green (B) SeeSupplementary Data Fig S2 for other spectral colour categories Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploidhybrid taxa Abbreviations p pink syn synthetic g green (C) Colour hexagon displaying the distribution of Nicotiana colour loci in bee colour space The hexa-gon has been scaled so that vertices represent 40 excitation of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-GUVndashgreen Bee colour categories are delineated by coloured ovals sat saturated Nicotiana species abbreviations are as follows acum acuminata aren arentsiiatten attenuata benavid benavidesii benth benthamiana clev clevelandii forst forsteri glau25 glauca 51725 glau51y glauca 51751 yellow glau51g glauca51751 green glut glutinosa goss gossei knight knightiana langs langsdorffii lin9647 linearis 964750099 linTW77 linearis TW77 mega megalosiphon miermiersii mut1w mutabilis CPG12456 white mut1p mutabilis CPG12456 pink mut3w mutabilis CPG3 white mut3p mutabilis CPG3 pink neso nesophila noctnoctiflora nudi nudicaulis obtus1 obtusiata line 1 obtus2 obtusiata line 2 obtus5 obtusiata line 5 obtusB obtusifolia var obtusifolia lsquoBaldwinrsquoobtusTW obtusifolia var obtusifolia TW143 obtuspalm obtusifolia var palmeri occhesp occidentalis subsp hesperis otoph w otophora white otoph p otophorapink pani paniculata pauc pauciflora petun petunioides plumba plumbaginifolia quad9047 quadrivalvis 904750042 quadTW18 quadrivalvis TW18 raimraimondii repa repanda rustasi rustica var asiatica rustpav rustica var pavonii syn UP synthetic UP syn F1 synthetic PUE1 F1 synrust S0 synthetic rus-tica PUE1-R10 S0 synrust S1 synthetic rustica PUE1-R1 S1 setch setchellii stock stocktonii suav suaveolens sylv6898 sylvestris 6898 sylvA047 sylvestrisA04750326 tab09555 tabacum 095-55 tab51789 tabacum 51789 tabchulu tabacum lsquoChulumanirsquo syntabQM synthetic tabacum QM syntabTH37 synthetictabacum TH37 tomtform tomentosiformis undu undulata wigan wigandioides TH32 TH32 synthetic N sylvestrisN otophora polyploid (D) Distribution ofNicotiana spectral loci in hummingbird colour space Vertices of the hummingbird colour space represent 50 excitation of the photoreceptors single photoreceptortype vertices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows repre-sent the vectors of these photoreceptors from the origin of the hummingbird colour space Nicotiana loci are coloured according to hummingbird colour categories(Supplementary Data Fig S3B) but are labelled with the accession name if the category includes only one taxon See Supplementary Data Video for an animation
of this graph
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1125
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
is the exception because it falls into an unexpected humming-bird colour category given the progenitor species (Fig 6)Section Repandae polyploids seem to have evolved to be eitherlike their maternal (Nicotiana nesophila N repanda andN stocktonii) or paternal (N nudicaulis) progenitor (Fig 6)The maternal progenitor N sylvestris is no longer sympatricwith any of the section Repandae polyploids therefore Nnesophila N repanda and N stocktonii can occupy the samepollination niche as their maternal progenitor without competi-tion Similarly section Suaveolentes is native to Australasiaexcept for one species in Namibia Africa and is not sympatricwith its progenitor sections in South America (Goodspeed1954) these polyploids and their diploid progenitors displaysimilar floral colours except N pauciflora which evolvedspectrally UVndashwhite flowers after the formation of sectionSuaveolentes (Fig 6) It is possible that there is less competitionfor pollinators and therefore reduced selective pressure to-wards floral colour diversification when polyploid species arenot sympatric with their diploid progenitors as is seen in
Iochrominae (Solanaceae) which have a broader range of floralcolours when species are sympatric (Muchhala et al 2014)However floral colour evolution can also be driven by geneticdrift or selection on pleiotropic effects of floral genes (Chittkaet al 2001 Rausher 2008) Furthermore anthocyanins andflavonoids are important for UV protection can defend againstfungi act as signal molecules and play a role in male fertility insome species in addition to their roles in signalling to pollina-tors (Shirley 1996)Over half of polyploids have an unexpected inheritance pat-
tern for chlorophyll if it is assumed that the presence of chloro-phyll is a dominant character Those polyploids that deviatefrom expectation span an age range of synthetic to 10myoand always lack chlorophyll in their petals which is in linewith the direction of the shifts observed in the divergence ofdiploid species (Supplementary Data Fig S6) In carnation thedifference in chlorophyll concentration between white andgreen flowers is likely caused by downregulation of chlorophyllbiosynthesis genes in white flowers genes involved in
TABLE 1 Polyploid and homoploid hybrid observed and expected floral colours
Species Spectral Bee Hummingbird Chlorophyll
Observed Expected Observed Expected Observed Expected Observed Expected
N tabacum 095-55 R W P UV-B B-G sUV-P W P N CN tabacum 51789 P W P B-G B-G P W P N CN tabacum lsquoChulumanirsquo W W P B-G B-G P W P N Csyn N tabacum QM P W P B B-G LP W P N Csyn N tabacum TH37 P W P B-G B-G P W P N CTH32 P W P B-G B-G P W P N CN rustica var asiatica G G LG G B-G G G W C CN rustica var pavonii G G G G B-G G G W C Csyn UP (homoploid) G G UV-G G B-G UV-G G W C Csyn F1 (homoploid) G G LG G B-G G G W C Csyn N rustica S0 G G LG G B-G G G W C Csyn N rustica S1 G G LG G B-G G G W C CN arentsii W G W B-G B-G P W C CN clevelandii W UV-W W B-G UV B-G P UV-W W N CN quadrivalvis TW18 W UV-W W B-G UV B-G W UV-W W C CN quadrivalvis 9047 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 1 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 2 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 5 W UV-W W B-G UV B-G W UV-W W C CN repanda W W UV-W B-G B-G UV W W UV-W N CN nesophila W W UV-W B-G B-G UV W W UV-W N CN stocktonii W W UV-W B-G B-G UV W W UV-W C CN nudicaulis UV-W W UV-W UV B-G UV UV-W W UV-W C CN benthamiana W W ndash ndash ndash ndash N CN forsteri W W ndash ndash ndash ndash C CN gossei W W ndash ndash ndash ndash N CN megalosiphon W W ndash ndash ndash ndash N CN occidentalis W W ndash ndash ndash ndash N CN suaveolens W W ndash ndash ndash ndash N C
N glauca 51725 Y W ndash ndash ndash ndash C CN glauca 51751 YG W ndash ndash ndash ndash C CN linearis TW77 UV-W W ndash ndash ndash ndash C CN linearis 9647 UV-W W ndash ndash ndash ndash C CN glutinosa P P W G ndash ndash ndash ndash C C
The top block consists of polyploid accessions and the bottom block includes homoploid hybridsItalic denotes an unexpected phenotype given the colour categories of the progenitorsProgenitor bee and hummingbird colour categories are unknown for section Suaveolentes and natural homoploid hybrids (see text)R red W white P pink G green UV-W UVndashwhite Y yellow UV-B UVndashblue B-G bluendashgreen B blue LG light green UV high UV UV-G
UVndashgreen sUV-P saturated UVndashpink LP light pink N no chlorophyll C chlorophyll syn synthetic 9047 represents 904750042 9647 represents964750099
1126 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
is the exception because it falls into an unexpected humming-bird colour category given the progenitor species (Fig 6)Section Repandae polyploids seem to have evolved to be eitherlike their maternal (Nicotiana nesophila N repanda andN stocktonii) or paternal (N nudicaulis) progenitor (Fig 6)The maternal progenitor N sylvestris is no longer sympatricwith any of the section Repandae polyploids therefore Nnesophila N repanda and N stocktonii can occupy the samepollination niche as their maternal progenitor without competi-tion Similarly section Suaveolentes is native to Australasiaexcept for one species in Namibia Africa and is not sympatricwith its progenitor sections in South America (Goodspeed1954) these polyploids and their diploid progenitors displaysimilar floral colours except N pauciflora which evolvedspectrally UVndashwhite flowers after the formation of sectionSuaveolentes (Fig 6) It is possible that there is less competitionfor pollinators and therefore reduced selective pressure to-wards floral colour diversification when polyploid species arenot sympatric with their diploid progenitors as is seen in
Iochrominae (Solanaceae) which have a broader range of floralcolours when species are sympatric (Muchhala et al 2014)However floral colour evolution can also be driven by geneticdrift or selection on pleiotropic effects of floral genes (Chittkaet al 2001 Rausher 2008) Furthermore anthocyanins andflavonoids are important for UV protection can defend againstfungi act as signal molecules and play a role in male fertility insome species in addition to their roles in signalling to pollina-tors (Shirley 1996)Over half of polyploids have an unexpected inheritance pat-
tern for chlorophyll if it is assumed that the presence of chloro-phyll is a dominant character Those polyploids that deviatefrom expectation span an age range of synthetic to 10myoand always lack chlorophyll in their petals which is in linewith the direction of the shifts observed in the divergence ofdiploid species (Supplementary Data Fig S6) In carnation thedifference in chlorophyll concentration between white andgreen flowers is likely caused by downregulation of chlorophyllbiosynthesis genes in white flowers genes involved in
TABLE 1 Polyploid and homoploid hybrid observed and expected floral colours
Species Spectral Bee Hummingbird Chlorophyll
Observed Expected Observed Expected Observed Expected Observed Expected
N tabacum 095-55 R W P UV-B B-G sUV-P W P N CN tabacum 51789 P W P B-G B-G P W P N CN tabacum lsquoChulumanirsquo W W P B-G B-G P W P N Csyn N tabacum QM P W P B B-G LP W P N Csyn N tabacum TH37 P W P B-G B-G P W P N CTH32 P W P B-G B-G P W P N CN rustica var asiatica G G LG G B-G G G W C CN rustica var pavonii G G G G B-G G G W C Csyn UP (homoploid) G G UV-G G B-G UV-G G W C Csyn F1 (homoploid) G G LG G B-G G G W C Csyn N rustica S0 G G LG G B-G G G W C Csyn N rustica S1 G G LG G B-G G G W C CN arentsii W G W B-G B-G P W C CN clevelandii W UV-W W B-G UV B-G P UV-W W N CN quadrivalvis TW18 W UV-W W B-G UV B-G W UV-W W C CN quadrivalvis 9047 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 1 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 2 W UV-W W B-G UV B-G W UV-W W C CN obtusiata line 5 W UV-W W B-G UV B-G W UV-W W C CN repanda W W UV-W B-G B-G UV W W UV-W N CN nesophila W W UV-W B-G B-G UV W W UV-W N CN stocktonii W W UV-W B-G B-G UV W W UV-W C CN nudicaulis UV-W W UV-W UV B-G UV UV-W W UV-W C CN benthamiana W W ndash ndash ndash ndash N CN forsteri W W ndash ndash ndash ndash C CN gossei W W ndash ndash ndash ndash N CN megalosiphon W W ndash ndash ndash ndash N CN occidentalis W W ndash ndash ndash ndash N CN suaveolens W W ndash ndash ndash ndash N C
N glauca 51725 Y W ndash ndash ndash ndash C CN glauca 51751 YG W ndash ndash ndash ndash C CN linearis TW77 UV-W W ndash ndash ndash ndash C CN linearis 9647 UV-W W ndash ndash ndash ndash C CN glutinosa P P W G ndash ndash ndash ndash C C
The top block consists of polyploid accessions and the bottom block includes homoploid hybridsItalic denotes an unexpected phenotype given the colour categories of the progenitorsProgenitor bee and hummingbird colour categories are unknown for section Suaveolentes and natural homoploid hybrids (see text)R red W white P pink G green UV-W UVndashwhite Y yellow UV-B UVndashblue B-G bluendashgreen B blue LG light green UV high UV UV-G
UVndashgreen sUV-P saturated UVndashpink LP light pink N no chlorophyll C chlorophyll syn synthetic 9047 represents 904750042 9647 represents964750099
1126 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
G 50
R 50
B 50
UV 50
R 25
B 25
UV 25
G 25
G 25
R 25
B 25 UV 25
0
0middot2
0middot4
0middot6
0middot8
1middot0
1middot0
1middot0
Reflecta
nce
Wavelength (nm)
N tabacum lt0middot2 myo
N tabacum lsquoChulumanirsquo
synthetic N tabacum QM
synthetic N tabacum TH37
N tomentosiformis
N sylvestris 6898
N sylvestris A04750326
N tabacum 095-55
N tabacum 51789
A
B C
B
B-G
G
UV-G
UV
UV-B
0
0middot2
0middot4
0middot6
0middot8
Reflecta
nce
Wavelength (nm)
N sylvestris 6898
N sylvestris A04750326
N nesophila
N nudicaulis
N repanda
N stocktonii
N obtusifolia var obtusifolia lsquoBaldwinrsquo
N obtusifolia var obtusifolia TW143
D
E F
B
B-G
G
UV-G
UV
UV-B
Repandae approx4middot5 myo
0
0middot2
0middot4
0middot6
0middot8
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
300 350 400 450 500 550 600 650 700
Reflecta
nce
Wavelength (nm)
B
B-G
G
UV-G
UV
UV-B
N noctiflora
N petunioides
N glauca 51725
N glauca 51751 yellow
N glauca 51751 green
Noctiflorae
N linearis 964750099
N linearis TW77
N acuminata
N attenuata
N miersii
N pauciflora
Petunioides
G
H INoctifloraendashPetunioides homoploids
y
z
x
y
z
x
y
z
x
FIG 5 (A D G) Reflectance spectra for polyploid and homoploid sections and their progenitors (A) N tabacum (D) section Repandae and (G)NoctifloraendashPetunioides homoploid hybrids Solid lines are used for diploid taxa dashed lines for polyploid taxa and dotted lines for homoploid hybrid taxa (B EH) Hummingbird colour space for polyploid and homoploid sections and their progenitors (B) N tabacum (E) section Repandae and (H) NoctifloraendashPetunioideshomoploid hybrids The vertices of the hummingbird colour space represent 25 (B E) or 50 (H) excitation of the photoreceptors single photoreceptor type ver-tices (red green blue and UV) are coloured red green blue and black respectively and all other vertices are grey Red green blue and black arrows represent thevectors of these photoreceptors from the origin of the hummingbird colour space (C F I) Bee colour hexagons for polyploid or homoploid sections and their progen-itors (C) N tabacum (F) section Repandae and (I) NoctifloraendashPetunioides homoploid hybrids Hexagons have been scaled so that vertices represent 40 excita-tion of photoreceptors UV ultraviolet UV-B UVndashblue B blue B-G bluendashgreen G green UV-G UVndashgreen For information regarding how to interpret colourhexagons see Supplementary Data Fig S1 Female ($) and male () symbols mark maternal and paternal progenitors respectively in the hummingbird and bee col-
our spaces
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1127
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
chlorophyll degradation are equally expressed in both flowertypes (Ohmiya et al 2014) It is possible that it is advantageousto limit the costs of chlorophyll production when it is unneces-sary for photosynthesis resulting in selection against the pres-ence of chlorophyll in petal tissue It is also possible that thisphenotype results from the silencing of the homeologues thatpromote chlorophyll biosynthesis
Transgressive flower colour in N tabacum and the syntheticpolyploid TH32
Polyploids N tabacum and synthetic TH32 are similar be-cause they share a maternal progenitor N sylvestris and theirpaternal progenitors Nicotiana tomentosiformis and N oto-phora respectively are both from section Tomentosae andhave similar reflectance spectra (Supplementary Data FigS5G) the paternal progenitors possess anthocyanin pigmenta-tion as well as chlorophyll whereas the maternal progenitorlacks both of theseGenetic crosses in Nicotiana suggest that both green flower
colour and the ability to produce floral anthocyanins are domi-nant and each may be determined by a single (likely multi-genic) locus (Brieger 1935) Nicotiana tabacum accessions andTH32 possess anthocyanin pigmentation (two spectral peaks inthe blue and red portions of the spectrum) but not chlorophyll(the lack of a reflectance minimum at 675 nm) as well as spec-tral reflectance curve shapes that are distinct from those of theirprogenitors (Fig 5A Supplementary Data Fig S4A)Therefore N tabacum and TH32 inherit anthocyanin floral pig-mentation from their paternal progenitors but a plastid pheno-type (chlorophyll is only found in plastids) like that of theirmaternal progenitor which likely has colourless leucoplasts asis seen in Arabidopsis (Pyke and Page 1998) Intriguinglyboth the N tomentosiformis and N sylvestris copies of thebHLH transcription factor involved in regulation of the antho-cyanin biosynthetic pathway are expressed and functional in Ntabacum (Bai et al 2011) suggesting that a maternal gene hasbeen co-opted into producing a paternal-type phenotypePolyploids typically inherit plastids from their maternal pro-
genitor it may be unsurprising therefore that N tabacum andTH32 plastids have the maternal phenotype However it islikely that the chloroplast-to-leucoplast transition in petal devel-opment is regulated by nuclear genes A study in Arabidopsisindicated that petal homeotic genes APETALA3 andPISTILLATA downregulate BANQUO genes which are in-volved in accumulation of chlorophyll suggesting that thebreakdown of chloroplasts in petal development is linked to re-pression of genes involved in chlorophyll biosynthesis by
Wh
ite
Pin
kR
ed
Ye
llow
Gre
en
Da
rk g
ree
n
Eq
uiv
oca
lN
od
e a
bse
nt
Spectr
al
UV
-wh
ite
Gre
en
Da
rk g
ree
n
Sat g
reen
UV
-gre
en
Sat U
V-g
reen
Hig
h U
V
Be
e
UV
-blu
e
Blu
e
Blu
e-g
ree
n
Lig
ht g
ree
nS
at U
V-p
ink
UV
-wh
ite
UV
-gre
en
Gre
en
Da
rk g
ree
nS
at
gre
en
Hum
min
gbird
Wh
ite
Pin
kLig
ht pin
k
N to
mento
siform
is
N oto
phora
N
ob
tusifo
lia v
ar
palm
eri
N
ob
tusifo
lia v
ar
ob
tusifo
lia
N m
iers
ii
N pauciflo
ra
N acum
inata
N attenuata
N sylv
estr
is
N plu
mbagin
ifolia
N la
ngsdorf
fii
N noctiflora
N petu
nio
ides
N w
igandio
ides
N undula
ta
N benavid
esii
N ra
imondii
N panic
ula
ta
N knig
htiana
1middot0
0middot9
9
1middot0
1middot0
1middot0 1middot0
1middot0
1middot0
1middot0
1middot01middot0
1middot0
0middot9
7
1middot01middot0
Spec
BeeH
um
N ta
bacum
N are
nts
ii
N ru
stica
N quadrivalv
is
N cle
vela
nd
ii
Repandae
N nudic
aulis
N re
panda
N sto
ckto
nii
N nesophila
Poly
dic
liae
N fo
rste
ri
N suaveole
ns
N gossei
N m
egalo
sip
hon
Su
ave
ole
nte
s
N glu
tinosa
N gla
uca
N lin
earis
lt0middot2
my
o
1 m
yo
4middot5
my
o
10 m
yo
Ho
mo
plo
id
hyb
rid
s
Tom
ento
sae
Trigonop
hylla
e
Petu
nio
ides
Sylv
estr
es
Ala
tae
No
ctiflo
rae
Undula
tae
Panic
ula
tae
0middot9
71
middot01
middot0
0middot9
8
1middot0
1middot0
FIG6Resultsofancestralstatereconstructionforspectralcolourcategories
summarized
onthe95
majority
ruletree
from
theBayesiananalysisofplastid
sequence
datafrom
non-hybriddiploidsPosterior
probabilitiesareshownbelowthebranchesHomoploid
andpolyploid
hybridsaresuperim
posedonthediploid
treeblack
andgreysoliddashed
anddotted
lines
totherightofthetree
representhybridization
eventsOrangebranches
wereadded
tothetree
whereprogenitorsofthehybridtaxaareentiresectionsPiechartsatinternalnodes
indicatecharacterstates
inferred
forthatnodeduringancestralstatereconstruc-
tioncarriedoutonasetof36000post-burn-intreesfrom
theBayesiananalysesPiechartsatthetipsofthebranches
indicatecharacterstates
observed
inextantspeciesBee
andhummingbirdcolourcategories
forextantspeciesaredisplayed
atthetipsoftheplastid
tree
TABLE 2 Mantel test results
Trait Geneticdistance
Bayesian
P value Mean P value significanttrees
Spectral reflectance 00229 002066 00215 901Bee colour vision 00866 004106 00321 662Hummingbird colour vision 00594 001986 00187 932
1128 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
nuclear-encoded petal identity genes (Mara et al 2010)Crosses in carnation and Nicotiana provide evidence that ma-ternal plastid phenotype does not determine that of its offspring(Brieger 1935 Ohmiya et al 2014) affirming that the plastidphenotype seen in N tabacum and TH32 polyploids isunexpected
Because this floral phenotype is unlike either progenitor anddivergent from the expected phenotype (ie the presence ofboth chlorophyll and anthocyanin pigments from their paternalprogenitor) and because the phenotype is seen in synthetic poly-ploids it can be considered to be caused by polyploidy and hy-bridization and is thus a transgressive phenotype Because allthree natural N tabacum accessions examined show the samephenotype as the synthetic polyploids we can infer that this un-expected floral phenotype is also transgressive in N tabacumThe observation of this phenotype in at least four independentorigins (three synthetic and the natural accessions) suggests thatthe interplay between the inheritance of plastid and vacuolarpigments yields a transgressive phenotype repeatedly in Ntabacum and TH32 polyploids
Nicotiana tabacum varies in spectral shape and bee andhummingbird colour categories among the accessions exam-ined here (Fig 5AndashC) Synthetic N tabacum QM and N taba-cum 095-55 are unexpected in both bee and hummingbirdcolour categories suggesting that these accessions will be dis-tinguishable from their progenitors by both bee (and likelyhawkmoth due to similarities in photoreceptor sensitivities)and hummingbird pollinators The differences seen among theN tabacum spectra may be due to the presence of differentcyanidin derivatives but vacuolar pH and the formation of het-erodimers of anthocyanin and flavonol pigments can also causeshifts in spectral reflectance (Grotewold 2006 Andersen andJordheim 2010)
Novel floral colour in homoploid hybrids
Over half of the homoploid hybrids examined show unex-pected phenotypes in spectral colour categories Without repro-ductive isolation homoploid hybrids often facilitate gene flowbetween their progenitors instead of becoming established asnew species (Buerkle et al 2000 2003) In experimental fieldplots of Nicotiana alata and Nicotiana forgetiana pollinator fi-delity decreased significantly in the presence of F1 hybrids in-creasing gene flow between the two progenitor species(Ippolito et al 2004) Homoploid hybrid N glauca displays anovel floral colour in spectral bee and hummingbird colour cat-egories (Fig 6) Although it is the combination of the suite offloral traits displayed that will influence what behaviour a polli-nator exhibits this change in floral colour may have played atleast some role in the establishment of reproductive isolationbetween N glauca and its progenitors
Species of progenitor sections Noctiflorae and Petunioidesmostly have vespertine flowers and many have long corollatubes (Goodspeed 1954) which suggests pollination by noctur-nal hawkmoths The only studies examining pollination in anyof these species have confirmed that N attenuata is pollinatedby nocturnal hawkmoths but is also visited by hummingbirds(Aigner and Scott 2002 Kessler and Baldwin 2006)Nicotiana glauca is pollinated by hummingbirds in its native
range (Nattero and Cocucci 2007) Selection can still occur inthe presence of generalist pollination based on differences inpollinator assemblage (Gomez et al 2009) so the floral colourshift in N glauca accompanied by a shift in the predominantpollinator may have aided reproductive isolation and its estab-lishment as a new species Evolutionary shifts in characteristicsknown to affect pollinator preferences often occur together Ashift from insect to hummingbird pollination has occurred twicewithin Mimulus section Erythranthe (Phrymaceae) and redflowers exserted stamens and pistils and reflexed upper petals(characters associated with hummingbird pollination) seem tohave evolved at the same points on the phylogenetic tree as theshift in pollination (Beardsley et al 2003) In addition to a shiftto yellow flowers N glauca has a reduced floral limb the partof the corolla that opens (associated with hummingbird pollina-tion) compared with many species in its progenitor sectionssuggesting the possibility of hummingbird-mediated selectionon N glauca floral traits
Concluding remarks
Floral colour shifts in polyploid and homoploid hybrids mayoccur immediately after their formation perhaps as a conse-quence of novel cisndashtrans interactions between progenitorgenomes (Chen 2007) Using genomic studies to examineplantndashpollinator interactions will shed light on the complexinteractions involved in successful pollination and pollinator-mediated evolution (Clare et al 2013) Transgressive and un-expected floral colours may have aided hybrid speciation butpollination studies of hybrids and their progenitors are neededto make these conclusions Typically synthetic and youngpolyploids (lt02myo) have floral colour that is unexpectedconsidering the colour of their progenitors in the colourperception of at least one pollinator type Older polyploids(1ndash10myo) tend to have a floral colour similar to at least oneprogenitor perhaps due to the fact that the polyploids are nolonger sympatric with one or both progenitors andor becauseother floral traits were more important in the divergence fromtheir progenitors
SUPPLEMENTARY DATA
Supplementary data are available online at wwwaoboxford-journalsorg and consist of the following Table S1 Nicotianaaccessions used in the spectral reflectance dataset and in petalcell area measurements Table S2 floral colour characters forall Nicotiana species examined Table S3 Tukeyrsquos honest sig-nificance test results for cell areas Table S4 polyploid andhomoploid hybrid origins Figure S1 navigating the bee colourhexagon Figure S2 Nicotiana reflectance spectra from 300 to700 nm by spectral colour category Figure S3 dendrogramsbased on distance cluster analyses for bee and hummingbirdcolour categories Figure S4 reflectance spectra bee colourhexagons and hummingbird colour space for TH32 N rusticaand N arentsii Figure S5 reflectance spectra bee colour hexa-gons and hummingbird colour space for section Polydicliaesection Suaveolentes and N glutinosa Figure S6 ancestral statereconstruction of the presenceabsence of chlorophyll in petals
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1129
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Video animation of Nicotiana spectra in 3-D hummingbird col-our space
ACKNOWLEDGEMENTS
We thank Michael Chester for helpful comments on the manu-script This work was supported by the Natural EnvironmentResearch Council (NEC5119641 to ARL and MWC) theCzech Science Foundation [P5011310057S to AK] and theOverseas Research Students Awards Scheme to EWM
LITERATURE CITED
Aharoni A De Vos CHR Wein M et al 2001 The strawberry FaMYB1transcription factor suppresses anthocyanin and flavonol accumulation intransgenic tobacco Plant Journal 28 319ndash332
Aigner PA Scott PE 2002 Use and pollination of a hawkmoth plant Nicotianaattenuata by migrant hummingbirds Southwestern Naturalist 47 1ndash11
Andersen OM Jordheim M 2010 Chemistry of flavonoid-based colors inplants In L Mander H-W Liu eds Comprehensive natural products IIchemistry and biology Oxford Elsevier 547ndash614
Anssour S Krugel T Sharbel TF Saluz HP Bonaventure G Baldwin IT2009 Phenotypic genetic and genomic consequences of natural and syn-thetic polyploidization of Nicotiana attenuata and Nicotiana obtusifoliaAnnals of Botany 103 1207ndash1217
Arnold SEJ Faruq S Salvolainen V McOwen PW Chittka L 2010 FReDthe floral reflectance database ndash a webportal for analyses of flower colourPLoS One 5 e14287
Bai Y Pattanaik S Patra B Werkman JR Xie CH Yuan L 2011Flavonoid-related basic helix-loop-helix regulators NtAn1a and NtAn1b oftobacco have originated from two ancestors and are funcitonally activePlanta 234 363ndash375
Beardsley PM Yen A Olmstead RG 2003 AFLP phylogeny ofMimulus sec-tion Erythranthe and the evolution of hummingbird pollination Evolution57 1397ndash1410
Beaulieu JM Leitch IJ Patel S Pendharkar A Knight CA 2008 Genomesize is a strong predictor of cell size and stomatal density in angiospermsNew Phytologist 179 975ndash986
Bowmaker JK 1998 Evolution of colour vision in vertebrates Eye 12541ndash547
Bowmaker JK Dartnall HJA 1980Visual pigments of rods and cones in a hu-man retina Journal of Physiology 298 501ndash511
Bradshaw HD Schemske DW 2003 Allele substitution at a flower colour lo-cus produces a pollinator shift in monkeyflowers Nature 426 176ndash178
Brieger FG 1935Genetic analysis of the cross between the self-fertile Nicotianalangsdorffii and the self-sterileN sanderae Journal of Genetics 30 79ndash100
Briscoe AD Chittka L 2001 The evolution of color vision in insects AnnualReview of Entomology 46 471ndash510
Buerkle CA Morris RJ Asmussen MA Rieseberg LH 2000 The likelihoodof homoploid hybrid speciationHeredity 84 441ndash451
Buerkle CA Wolf DE Rieseberg LH 2003 The origin and extinction of spe-cies through hybridization In CA Brigham MWSchwartz eds Populationviability in plants conservation management and modeling of rare plantsNewYork Springer 117ndash141
Castaneda-Ovando A Pacheco-Hernandez ML Paez-Hernandez MERodriguez JA Galan-Vidal CA 2009 Chemical studies of anthocyaninsa review Food Chemistry 113 859ndash871
Chase MW Knapp S Cox AV et al 2003 Molecular systematics GISH andthe origin of hybrid taxa in Nicotiana (Solanaceae) Annals of Botany 92107ndash127
Chen ZJ 2007 Genetic and epigenetic mechanisms for gene expression andphenotypic variation in plant polyploids Annual Review of Plant Biology58 377ndash406
Chittka L 1992 The color hexagon a chromaticity diagram based on photore-ceptor excitations as a generalized representation of color opponencyJournal of Comparative Physiology A Sensory Neural and BehavioralPhysiology 170 533ndash543
Chittka L 1996 Optimal sets of colour receptors and opponent processes forcoding of natural objects in insect vision Journal of Theoretical Biology181 179ndash196
Chittka L Waser NM 1997Why red flowers are not invisible for bees IsraelJournal of Plant Sciences 45 169ndash183
Chittka L Spaethe J Schmidt A Hickelsberger A 2001 Adaptation con-straint and chance in the evolution of flower color and pollinator color vi-sion In L Chittka JD Thompson eds Cognitive ecology of pollinationCambridge Cambridge University Press 106ndash126
Clare EL Schiestl FP Leitch AR Chittka L 2013 The promise of genomicsin the study of plant-pollinator interactionsGenome Biology 14 207
Clarkson JJ Knapp S Garcia VF Olmstead RG Leitch AR Chase MW2004 Phylogenetic relationships in Nicotiana (Solanaceae) inferred frommultiple plastid DNA regions Molecular Phylogenetics and Evolution 3375ndash90
Clarkson JJ Lim KY Kovarik A Chase MW Knapp S Leitch AR 2005Long-term genome diploidization in allopolyploid Nicotiana sectionRepandae (Solanaceae) New Phytologist 168 241ndash252
Clarkson JJ Kelly LJ Leitch AR Knapp S Chase MW 2010Nuclear gluta-mine synthetase evolution in Nicotiana phylogenetics and the origins of al-lotetraploid and homoploid (diploid) hybrids Molecular Phylogenetics andEvolution 55 99ndash112
Cubo J Ponton F Laurin M de Margerie E Catanet J 2005 Phylogeneticsignal in bone microstructure of sauropsids Systematic Biology 54562ndash574
Gaeta RT Pires JC Iniguez-Luy F Leon E Osborn TC 2007 Genomicchanges in resynthesized Brassica napus and their effect on gene expressionand phenotype Plant Cell 19 3403ndash3417
Goldsmith TH 1980 Hummingbirds see near ultraviolet light Science 207786ndash788
Goldsmith TH Goldsmith KM 1979 Discrimination of colors by the black-chinned hummingbird Archilochus alexandri Journal of ComparativePhysiology 130 209ndash220
Gomez JM Perfectti F Bosch J Camacho JPM 2009A geographic selectionmosaic in a generalized plant-pollinator-herbivore system EcologicalMonographs 79 245ndash263
Goodspeed TH 1954 The genus NicotianaWaltham MA Chronica BotanicaGrant V 1952 Isolation and hybridization between Aquilegia formosa and A
pubescensAliso 2 341ndash360Grotewold E 2006 The genetics and biochemistry of floral pigments Annual
Review of Plant Biology 57 761ndash780Gumbert A Kunze J Chittka L 1999 Floral colour diversity in plant commu-
nities bee colour space and a null model Proceedings of the Royal SocietyB Biological Sciences 266 1711ndash1716
Haardt HMaske H 1987 Specific in vivo absorption-coefficient of chlorophylla at 675 nm Limnology and Oceanography 32 608ndash619
Hart NS Hunt DM 2007 Avian visual pigments characteristics spectral tun-ing and evolution American Naturalist 169 S7ndashS26
Herrera G Zagal JC DiazM et al 2008 Spectral sensitivities of photoreceptorsand their role in colour discrimination in the green-backed firecrown hum-mingbird (Sephanoides sephanoides) Journal of Comparative Physiology ANeuroethology Sensory Neural and Behavioral Physiology 194 785ndash794
Hoballah ME Gubitz T Stuurman J et al 2007 Single gene-mediated shiftin pollinator attraction in Petunia Plant Cell 19 779ndash790
Ippolito A Fernandes GW Holtsford TP 2004 Pollinator preferences forNicotiana alata N forgetiana and their F1 hybrids Evolution 582634ndash2644
Kaczorowski RL Gardener MC Holtsford TP 2005 Nectar traits inNicotiana section Alatae (Solanaceae) in relation to floral traits pollinatorsand mating system American Journal of Botany 92 1270ndash1283
Kelber A 2001 Receptor based models for spontaneous colour choices in fliesand butterflies Entomologia Experimentalis et Applicata 99 231ndash244
Kelber A Balkenius AWarrant EJ 2003Colour vision in diurnal and noctur-nal hawkmoths Integrative and Comparative Biology 43 571ndash579
Kelly LJ Leitch AR Clarkson JJ Hunter RB Knapp S Chase MW 2010Intragenic recombination events and evidence for hybrid speciation inNicotiana (Solanaceae)Molecular Biology and Evolution 27 781ndash799
Kelly LJ Leitch AR Clarkson JJ Knapp S Chase MW 2013Reconstructing the complex origin of wild allotetraploid tobaccos(Nicotiana section Suaveolentes) Evolution 67 80ndash94
Kessler D Baldwin IT 2006Making sense of nectar scents the effects of nec-tar secondary metabolites on floral visitors of Nicotiana attenuata PlantJournal 49 840ndash854
Kevan P Giurfa M Chittka L 1996Why are there so many and so few whiteflowers Trends in Plant Science 1 280ndash284
1130 McCarthy et al mdash Floral colour evolution in Nicotiana polyploids
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from
Knapp S 2010 On lsquovarious contrivancesrsquo pollination phylogeny and flowerform in the Solanaceae Philosophical Transactions of the Royal Society BBiological Sciences 365 449ndash460
Ladiges PY Marks CE Nelson G 2011 Biogeography of Nicotiana sectionSuaveolentes (Solanaceae) reveals geographical tracks in arid AustraliaJournal of Biogeography 38 2066ndash2077
Leitch AR Leitch IJ 2008 Genomic plasticity and the diversity of polyploidplants Science 320 481ndash483
Leitch IJ Bennett MD 2004 Genome downsizing in polyploid plantsBiological Journal of the Linnean Society 82 651ndash663
Leitch IJ Hanson L Lim KY et al 2008 The ups and downs of genome sizeevolution in polyploid species of Nicotiana (Solanaceae) Annals of Botany101 805ndash814
Loytynoja A Goldman N 2008 Phylogeny-aware gap placement prevents er-rors in sequence alignment and evolutionary analysis Science 3201632ndash1635
MaddisonWPMaddison DR 2008Mesquite a modular system for evolution-ary analysis Version 25 httpmesquiteprojectorg Last accessed March31 2015
Mara CD Huang TB Irish VF 2010 The Arabidopsis floral homeotic proteinsAPETALA3 and PISTILLATA negatively regulate the BANQUO genes im-plicated in light signaling Plant Cell 22 690ndash702
McClintock B 1984 The significance of responses of the genome to challengeScience 226 792ndash801
Menzel R Ventura DF Hertel H de Souza JM Greggers U 1986Spectral sensitivity of photoreceptors in insect compound eyes Comparisonof species and methods Journal of Comparative Physiology A 158165ndash177
Muchhala N Johnsen S Smith SD 2014 Competition for hummingbirdpollination shapes flower color variation in Andean Solanaceae Evolution68 2275ndash2286
Nattero J Cocucci AA 2007 Geographical variation in floral traits of the treetobacco in relation to its hummingbird pollinator fauna Biological Journalof the Linnean Society 90 657ndash667
Ohmiya A Hirashima M Yagi M Tanase K Yamamizo C 2014Identification of genes associated with chlorophyll accumulation in flowerpetals PLOS One 9 e113738
Oksanen J Blanchet FG Kindt R et al 2013 vegan Community EcologyPackage 20-10 ed httpcranr-projectorgwebpackagesveganindexhtml Last accessed March 31 2015
Paradis E Claude J Strimmer K 2004 APE analyses of phylogenetics andevolution in R language Bioinformatics 20 289ndash290
Peitsch D Fietz A Hertel H Desouza J Ventura DF Menzel R 1992 Thespectral input systems of hymenopteran insects and their receptor-based
colour vision Journal of Comparative Physiology A Sensory Neural andBehavioral Physiology 170 23ndash40
Pyke KA Page AM 1998 Plastid ontogeny during petal development inArabidopsis Plant Physiology 116 797ndash803
Raine NE Ings TC Dornhausz A Saleh N Chittka L 2006 Adaptation ge-netic drift pleiotropy and history in the evolution of bee foraging behaviorAdvances in the Study of Behavior 36 305ndash354
Rausher MD 2008 Evolutionary transitions in floral color InternationalJournal of Plant Sciences 169 7ndash21
Restrepo A 2013 Hue processing in tetrachromatic spaces In KOEgiazarian SS Agaian AP Gotchev eds Image processing algorithmsand systems XI February 2013 Burlingame CA SPIE doi101117122003216
Rodriguez-Girones MA Santamaria L 2004 Why are so many bird flowersred PLoS Biology 2 e350
Shirley BW 1996 Flavonoid biosynthesis lsquonewrsquo functions for an lsquooldrsquo pathwayTrends in Plant Science 1 377ndash382
Shrestha M Dyer AG Boyd-Gerny S Wong BB Burd M 2013 Shades ofred bird-pollinated flowers target the specific colour discrimination abilitiesof avian vision New Phytologist 198 301ndash310
Skorupski P Doring TF Chittka L 2007 Photoreceptor spectral sensitivity inisland and mainland populations of the bumblebee Bombus terrestrisJournal of Comparative Physiology A Neuroethology Sensory Neuraland Behavioral Physiology 193 485ndash494
Soltis DE Albert VA Leebens-Mack J et al 2009 Polyploidy and angiospermdiversification American Journal of Botany 96 336ndash348
Soltis DE Segovia-Salcedo MC Jordon-Thaden I et al 2014 Are polyploidsreally evolutionary dead-ends (again) A critical reappraisal of Mayroseet al (2011)New Phytologist 202 1105ndash1117
Tirosh I Reikhav S Levy AA Barkai N 2009 A yeast hybrid providesinsight into the evolution of gene expression regulation Science 324659ndash662
Waser NM Chittka L Price MV Williams NM Ollerton J 1996Generalization in pollination systems and why it matters Ecology 771043ndash1060
Whittall JB Hodges SA 2007 Pollinator shifts drive increasingly long nectarspurs in columbine flowersNature 447 706ndash709
Wittkopp PJ Haerum BK Clark AG 2004 Evolutionary changes in cis andtrans gene regulationNature 430 85ndash88
Wyszecki G Stiles WS 1982 Color science concepts and methods quantita-tive data and formulaeNewYorkWiley
Zhu C Gerjets T Sandmann G 2007 Nicotiana glauca engineered for theproduction of ketocarotenoids in flowers and leaves by expressing the cya-nobacterial crtO ketolase gene Transgenic Research 16 813ndash821
McCarthy et al mdash Floral colour evolution in Nicotiana polyploids 1131
at University of G
reenwich on June 16 2015
httpaoboxfordjournalsorgD
ownloaded from